THE ACOUSTIC WAVE SENSOR AND SOFT LITHOGRAPHY …
174
THE ACOUSTIC WAVE SENSOR AND SOFT LITHOGRAPHY TECHNOLOGIES FOR CELL BIOLOGICAL STUDIES by Fang Li BS, Tsinghua University, P.R. China, 1999 MS, Tsinghua University, P.R. China, 2002 MS, University of Pittsburgh, 2004 Submitted to the Graduate Faculty of the Swanson School of Engineering in partial fulfillment of the requirements for the degree of Doctor of Philosophy University of Pittsburgh 2008
Transcript of THE ACOUSTIC WAVE SENSOR AND SOFT LITHOGRAPHY …
THE ACOUSTIC WAVE SENSOR AND SOFT LITHOGRAPHY TECHNOLOGIES FOR CELL BIOLOGICAL STUDIES
by
Fang Li
BS, Tsinghua University, P.R. China, 1999
MS, Tsinghua University, P.R. China, 2002
MS, University of Pittsburgh, 2004
Submitted to the Graduate Faculty of
the Swanson School of Engineering in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
University of Pittsburgh
2008
UNIVERSITY OF PITTSBURGH
SWANSON SCHOOL OF ENGINEERING
This dissertation was presented
by
Fang Li
It was defended on
May 27, 2008
and approved by
Co-Director: Dr. James H.-C. Wang, Associate Professor, Orthopeadic Surgery and Mechanical Engineering and Material Science
Dr. William S. Slaughter, Associate Professor, Mechanical Engineering and Material Science
Dr. Patrick Smolinski, Associate Professor, Mechanical Engineering and Material Science
Dissertation Director: Dr. Qing-Ming Wang, Associate Professor, Mechanical Engineering and Material Science
ii
Copyright © by Fang Li
2008
iii
THE ACOUSTIC WAVE SENSOR AND SOFT LITHOGRAPHY TECHNOLOGIES FOR CELL BIOLOGICAL STUDIES
Fang Li, PhD
University of Pittsburgh, 2008
Recently, cell-based biosensors have attracted many attentions because of their potential
applications in fundamental biological research, drug development, and other fields. Acoustic
wave biosensors offer powerful tools to probe cell behaviors and properties in a non-invasive,
simple, and quantitative manner. Current studies on cell-based acoustic wave sensors are focused
on experimental investigation of thickness shear mode (TSM) sensors for monitoring cell
attachment and spreading. There are no theoretical models for cell-based TSM biosensors. No
studies on other cell biological applications of TSM sensors or on surface acoustic wave cell-
based biosensors have been performed. The reliability and sensitivity of current cell-based
biosensors are low. Improving them requires studies on engineering cells and understanding the
effects of cell morphology on cell function.
The overall objective of this dissertation is to develop acoustic wave sensor systems for
cell biological studies and to determine the effects of cell shape on cell function. Our study
includes three parts: (1) Development of cell-based TSM sensor system; (2) Studies of Love
mode devices as cell-based biosensors; (3) Studies of the effects of cell shape on cell function. In
the first part, a theoretical model was developed, changes in cell adhesion were monitored and
cell viscoelasticity was characterized by TSM sensor systems. The TSM sensor systems were
demonstrated to provide a non-invasive, simple, and reliable method to monitor cell adhesion
and characterize cell viscoelasticity. In the second part, a theoretical model was developed to
iv
determine signal changes in Love mode sensors due to cells attaching on their surface.
Experimental results validated the model. In the third part, cell shape was patterned to different
aspect ratios. Elongated tendon cells were found to express higher collagen type I than shorter
cells. Changes in cell shape induced alterations in cytoskeleton, focal adhesions, and traction
forces in cells, which may collectively prompt the observed differential collagen type I
expression in cells with different shapes. Overall, our research expanded the applications of
acoustic wave cell-based biosensors. Studies on cell shape control and the effects of cell shape
on cell function will be useful for increasing the sensitivity of cell-based biosensors in future
research.
v
TABLE OF CONTENTS
PREFACE...................................................................................................................................XV
1.0 BACKGROUND........................................................................................................... 1
1.1 SENSORS, BIOSENSORS AND CELL-BASED BIOSENSORS................... 1
1.2 ACOUSTIC WAVE SENSORS.......................................................................... 4
1.2.1 General Background...................................................................................... 5
1.2.2 Classifications and Comparisons of Acoustic Wave Sensors ................... 12
1.2.3 Thickness Shear Mode (TSM) Devices....................................................... 16
1.2.4 Love-mode Acoustic Wave Sensors ............................................................ 23
1.3 MICROFABRICATION AND SOFT LITHGRPAHY ................................. 26
1.4 CELL BIOLOGICAL BACKGROUND......................................................... 32
1.4.1 Cellular Structure ........................................................................................ 32
1.4.2 Extracellular matrix (ECM)........................................................................ 38
1.4.3 Cell-ECM adhesion ...................................................................................... 39
1.4.4 Cell viscoelasticity ........................................................................................ 43
1.4.5 Cell traction force......................................................................................... 48
2.0 RESEARCH OBJECTIVE........................................................................................ 51
3.0 MONITORING CELL ATTACHMENT, SPREADING AND FORMATION OF FOCAL ADHESION BY USING THICKNESS SHEAR MODE ACOUSTIC WAVE SENSORS .................................................................................................................................... 53
3.1 INTRODUCTION ............................................................................................. 53
vi
3.2 THEORETICAL CONSIDERATIONS OF CELL-BASED TSM ACOUSTIC WAVE SENSOR........................................................................................... 55
3.2.1 Generic modeling of quartz TSM sensor with loading ............................. 55
3.2.2 Acoustic impedance of uniform multilayer loading and non-uniform loading ........................................................................................................................ 59
3.2.3 Physical model of cell adhesion onto the sensor surface and its acoustic loading impedance...................................................................................................... 61
3.3 MATERIALS AND METHODS...................................................................... 63
3.3.1 Experimental setup ...................................................................................... 63
3.3.2 Cell culture.................................................................................................... 65
3.3.3 Monitoring cell attachment, spreading and formation of focal adhesion65
3.3.4 Cells with cytochalasin D (CD) treatment ................................................. 66
3.3.5 Immunostaining of actin filaments and nucleus........................................ 66
3.4 RESULTS ........................................................................................................... 66
3.4.1 Modeling results for cell adhesion .............................................................. 66
3.4.2 Process of cell adhesion after cells contact with a surface........................ 68
3.4.3 Cell layers with CD treatment .................................................................... 70
3.5 DISCUSSIONS................................................................................................... 73
3.6 CONCLUSIONS................................................................................................ 75
4.0 THICKNESS SHEAR MODE RESONATORS FOR CHARACTERIZING VISCOELASTIC PROPERTIES OF MC3T3 FIBROBLASTS MONOLAYER................ 77
4.1 INTRODUCTION ............................................................................................. 77
4.2 MATERIALS AND METHODS...................................................................... 79
4.2.1 Experimental setup ...................................................................................... 79
4.2.2 Cell Culture................................................................................................... 79
4.2.3 Measurements............................................................................................... 80
vii
4.3 DETERMINATION OF THE VISCOELASTIC PROPERTIES OF CELL LAYERS ............................................................................................................................. 80
4.3.1 Generic consideration for determining the viscoelastic properties of cell monolayer.................................................................................................................... 81
4.3.2 Algorithmic strategy of exacting the viscoelastic parameters of cell layer . ........................................................................................................................ 82
4.4 RESULTS AND DISCUSSIONS...................................................................... 86
4.5 CONCLUSIONS................................................................................................ 94
5.0 LOVE MODE SURFACE ACOUSTIC WAVE DEVICE AS CELL-BASED BIOSENSOR ............................................................................................................................... 95
5.1 INTRODUCTION ............................................................................................. 95
5.2 THEORETICAL MODELING OF CELL-BASED LOVE MODE ACOUSTIC WAVE SENSORS ........................................................................................ 96
5.3 OPTIMIZATION OF LOVE MODE ACOUSTIC WAVE SENSOR DESIGN ........................................................................................................................... 102
5.4 EXPERIMENTAL CHARACTERIZATION OF LOVE MODE ACOUSTIC WAVE SENSORS ...................................................................................... 110
5.4.1 Fabrication of Love mode acoustic wave devices .................................... 110
5.4.2 Experimental setup .................................................................................... 113
5.4.3 Experimental results .................................................................................. 114
5.5 POTENTIAL APPLICATIONS OF LOVE MODE ACOUSTIC WAVE SENSOR FOR CELL BIOLOGICAL STUDIES ......................................................... 120
5.6 SUMMARY...................................................................................................... 125
6.0 THE APPLICATION OF SOFT LITHOGRAPHY TECHNIQUES TO DETERMINE THE EFFECTS OF CELL SHAPE ON CELL STRUCTURES AND FUNCTIONS............................................................................................................................. 127
6.1 INTRODUCTION ........................................................................................... 127
6.2 MATERIALS AND METHODS.................................................................... 129
6.2.1 Fabrication of micropatterned substrates ............................................... 129
6.2.2 Cell culture experiments............................................................................ 132
viii
6.2.3 Immunofluorescence assays for collagen type I, GAPDH, actin filaments, and vinculin............................................................................................................... 133
6.2.4 Cell traction force microscopy (CTFM)................................................... 133
6.2.5 Quantification of collagen type I expression levels ................................. 134
6.2.6 Statistical analysis ...................................................................................... 135
6.3 RESULTS ......................................................................................................... 135
6.3.1 Micropatterning human tendon fibroblasts with different shapes........ 135
6.3.2 Expression levels of collagen type I in micropatterned HTFs................ 136
6.3.3 Organization of actin filaments and focal adhesions in micropatterned HTFs ...................................................................................................................... 137
6.3.4 Cell traction forces in micropatterned HTFs .......................................... 138
6.4 DISCUSSION................................................................................................... 140
7.0 CONCLUDING REMAKS...................................................................................... 145
7.1 SUMMATY OF ACCOMPLISHMENTS..................................................... 146
7.2 FUTURE PERSPECTIVE.............................................................................. 148
BIBLIOGRAPHY..................................................................................................................... 150
ix
LIST OF TABLES
Table 1. Classification of acoustic wave sensors [10] ................................................................. 13
Table 2. Selected properties of acoustic wave sensors [10]......................................................... 15
Table 3 Surface mechanical impedance (Zs) for different surface perturbations ........................ 22
Table 4 Quartz parameters by sensor calibration ......................................................................... 86
Table 5 The extracted viscoelastic properties of cells for three individual experiments ............. 87
Table 6 Comparison of cell storage and viscosity with different techniques............................... 90
Table 7 Parameters of the materials for the simulation of Love mode devices ......................... 109
Table 8 Experimental conditions of PMMA spin coating and the thickness of PMMA film.... 115
x
LIST OF FIGURES
Figure 1. Principle of biosensors.................................................................................................... 3
Figure 2. Structure of bulk acoustic wave resonator...................................................................... 8
Figure 3. Interdigital transducer, formed by patterning electrodes on the surface of piezoelectric materials for the excitation of surface acoustic waves [2, 10]........................................................ 9
Figure 4. Geometry of two-port SAW devices: Two-port SAW resonator (A) and delay lines (B) [10]. ............................................................................................................................................... 10
Figure 5. Oscillation circuit of bulk acoustic wave (BAW) and surface acoustic wave (SAW) devices........................................................................................................................................... 11
Figure 6. Devices characterized with network analyzer by measuring forward-scattering parameter S21 (A) the reflection coefficient S11 (B) and a one-port resonator is characterized by impedance analyzer (C) [10]......................................................................................................... 12
Figure 7. Circular TSM quartz resonator (A). A quartz plate is cut with an angle of 35.10 degrees with respect to the optical z-axis. (B) [10]. ..................................................................... 17
Figure 8. Transmission line model to describe the near-resonant electrical characteristics of the resonator........................................................................................................................................ 18
Figure 9. Butterworth-van-Dyke (BVD) equivalent circuit of unperturbed TSM resonator near the series resonant frequency (A); Modified Butterworth-van-Dyke equivalent circuit of TSM resonators with small loadings (B) ............................................................................................... 20
Figure 10. Two surface-localized shear modes in anisotropic materials (SSBW (A) and Leaky SAW (B)) [10] .............................................................................................................................. 24
Figure 11. Love wave sensor. [10].............................................................................................. 25
Figure 12 Schematic procedures for microcontact printing of hexadecanethiol on the surface of gold [32]........................................................................................................................................ 28
Figure 13 Schematic illustration of procedures for (a) replica molding [34], (b) microtransfer molding (μTM), (c) micromolding in capillaries (MIMIC), and (d ) solvent-assisted micromolding (SAMIM)[32]. ....................................................................................................... 31
xi
Figure 14 Cell plasma membrane consisting of phospholipids and proteins [47]. ...................... 33
Figure 15. Three types of cytoskeleton: (a) actin filaments, (b) microtubule and (c) intermediate filaments........................................................................................................................................ 34
Figure 16. (A) Tensegrity model. (B) Force balance between tensed microfilaments (MFs), intermediate filaments (IFs), compressed microtubules (MTs) and the ECM in a region of a cellular tensegrity array. [49]. ....................................................................................................... 36
Figure 17. Focal contact or adhesion [47]................................................................................... 40
Figure 18. Immunofluorescence image of the focal adhesions [56]. ........................................... 40
Figure 19. Three states of cell-ECM adhesion [60]. .................................................................... 42
Figure 20. Principle of electric cell-substrate impedance sensing (ECIS) for monitoring cell adhesion [63]................................................................................................................................. 43
Figure 21. Schematic depiction of common experimental techniques for measuring cell visoelasticity [70].......................................................................................................................... 45
Figure 22. Transfer matrix model of a thickness shear mode (TSM) quartz resonator with multiple uniform layers loaded on one side.................................................................................. 57
Figure 23. Equivalent circuit of TSM resonator with non-uniform loading................................ 60
Figure 24. (a) cell monolayer adhering to TSM sensor surface; (b) multilayer model of cell monolayer on sensor surface......................................................................................................... 61
Figure 25. Non-uniform cell adhesion to TSM sensor surface .................................................... 62
Figure 26. Experimental setup of thickness shear mode resonators for cell-based functional biosensor ....................................................................................................................................... 64
Figure 27. Modeling results of relative motional resistance increase and resonant frequency decrease to cell-free medium due to cell monolayer adhesion. (A) Relative resistance increase with changing thickness of interfacial layer and viscosity of cell layer when the storage shear modulus of cell layer is fixed to 1KPa; (B) relative resistance increase with changing thickness of interfacial layer and storage shear modulus of cell layer when the viscosity of cell layer is fixed to 2cp; (C) relative frequency decrease with changing thickness of interfacial layer and viscosity of cell layer when the storage shear modulus of cell layer is fixed to 1KPa; and (D) relative frequency decrease with changing thickness of interfacial layer and storage shear modulus of cell layer when the viscosity of cell layer is fixed to 2cp. ......................................... 68
Figure 28. (A) Images of cells in culture at different time points; (B) time course of the relative motional resistance increase; and (C) relative resonant frequency decrease to the cell-free medium after different numbers of human skin fibroblasts seeded onto the resonator surface at time point zero .............................................................................................................................. 70
xii
Figure 29. Florescence images of human skin fibroblasts exposed to 5uM cytochalasin D for different time: (A) 0min; (B) 3min; (C) 5 min; (D) 10 min; (E) 20 min; (F) 40min; and (G) changes in motional resistance after confluent human skin fibroblast layers treated with Cytochalasin D with different concentrations............................................................................... 72
Figure 30. Flow chat for exacting the properties of loading layers from the measurements of the electric admittance spectra of TSM resonators............................................................................. 83
Figure 31. Comparison of G (real part of the electrical admittance) and B (imaginary part of the electrical admittance) spectra for the resonators with cell adhesion between the actual experimental measurements and calculated values by plugging the solutions into our theoretical model............................................................................................................................................. 88
Figure 32. The relationship between extracted values of the thickness of interfacial layer and storage shear modulus (A) and between extracted values of the thickness of interfacial layer and loss shear modulus (B).................................................................................................................. 89
Figure 33. (A) Models for Love mode acoustic wave sensors with multilayer loadings; (B) Transfer matrix model of a Love mode acoustic wave device with multiple uniform layers loaded on one side .................................................................................................................................... 97
Figure 34. (A) Propagation velocity v.s. the thickness of the guiding layer; (B) Propagation velocity v.s. the thickness of the guiding layer; (C) Phase shift in wave guiding layer v.s. the thickness of wave guiding layer; (D) mass sensitivity and phase shift in guiding layer v.s. the thickness of guiding layer. The materials parameters of substrate and guiding layer for each curve are (vs=5060m/s, vgl=707m/s, densitys=2648kg/m3, densitygl=1000kg/m3 for s1 and ϕ1; vs=5060m/s, vgl=1000m/s, densitys=2648kg/m3, densitygl=1000kg/m3 for s2 and ϕ2; vs=5060m/s, vgl=1414m/s, densitys=2648kg/m3, densitygl=1000kg/m3 for s3 and ϕ3; vs=4108m/s, vgl=1000m/s, densitys=2648kg/m3, densitygl=1000kg/m3 for s4 and ϕ4) .......................................................... 104
Figure 35. (A) Propagation velocity v.s. thickness of wave guiding layer; (B) Wave attenuation v.s. thickness of wave guiding layer; (C) Propagation velocity sensitivity to mass loading v.s. thickness of wave guiding layer; (D) Attenuation sensitivity to mass loading v.s. thickness of wave guiding layer...................................................................................................................... 108
Figure 36. Fabricated shear-horizontal acoustic wave devices.................................................. 111
Figure 37. Love mode acoustic wave devices............................................................................ 112
Figure 38. Love mode acoustic wave devices mounted on the test fixture ............................... 113
Figure 39. Measurement system of Love mode acoustic wave sensors..................................... 114
Figure 40. The amplitude and phase images of S21for Love mode acoustic sensors with 2.0mm PMMA coating............................................................................................................................ 116
Figure 41. Comparison of theoretical and experimental results of resonant frequency shift with the thickness of PMMA coating layer. ....................................................................................... 119
xiii
Figure 42. Comparison of theoretical and experimental results of insertion loss changes induced by PMMA layer with the different thickness.............................................................................. 120
Figure 43. Effect of the thickness of interfacial layer and wave guiding layer on the relative changes in (A) the propagation velocity and (B) attenuation to the semi-infinite cell medium loading......................................................................................................................................... 122
Figure 44 Effects of (A) storage shear modulus of cell layer, (B) viscosity of cell layer and (C) viscosity of interfacial layer on the relative changes in the propagation velocity and attenuation to the semi-infinite cell medium loading. ................................................................................... 124
Figure 45. A schematic of microcontact printing method used for micropatterning gold surfaces in this study. ................................................................................................................................ 131
Figure 46. A schematic of microcontact printing method used for micropatterning PA gel in this study............................................................................................................................................ 131
Figure 47. HTFs on (A) Micropatterned gold-coated substrates and (B) micropatterned polyacrylamide gel. (Scale bars: 50 μm). ................................................................................... 136
Figure 48. Immunofluorescence microscopy images of collagen type I expression in rectangular HTFs with aspect ratios of 19.6 (A), 4.9 (B), and 2.2 (C), respectively, and circular HTF (D). (Scale bars: 20 μm). .................................................................................................................... 137
Figure 49. Effect of cell shape on collagen type I expression level........................................... 138
Figure 50. Immunofluorescence microscopy images of F-actin, vinculin, and the overlay of them, as well as CTFs in micropatterned HTFs. A-C. Rectangular HTFs with aspect ratios of 19.6, 4.9, and 2.2, respectively; D circular HTF. (Scale bars: 20 μm). ...................................................... 139
Figure 51. The effect of cell shape on CTF magnitude. It increased with decreasing aspect ratio of the cell..................................................................................................................................... 140
xiv
PREFACE
During my study at the University of Pittsburgh, I have received a lot of support and help from
many people. I would like to take this opportunity to thank them.
First and foremost, I would like to thank my advisers, Prof. Qing-Ming Wang and Prof.
James H.-C. Wang, for offering me such a great interdisciplinary project and giving me a unique
training in the fields of mechanical engineering and cell biology. They always support and help
me in every possible way professors can help their students. From them, I not only learned how
to do excellent research, but also be a nice person. It is really a great experience to work with them.
Thank you very much! I would like to thank my other committee members, Prof. William S.
Slaughter and Prof. Patrick Smolinski, for their being on my committee, their support and all the
useful discussions. I would like to thank Dr. Minking K. Chyu for his support and advice in my job
search.
Special thanks to Dr. Bin Li in the mechanobiology laboratory for giving me so much
help and support on cell patterning project. Mr. Lifeng Qin helped me a lot in terms of the testing
of acoustic wave sensors. I appreciate it very much! I also would like to thank other members in
Microsensor and Microactuator Laboratory for their help and discussion regarding the
fabrication and testing of sensors and actuators and other members in Mechanobiology
Laboratory for giving me so much support in learning cell culture, RT-PCR, immunoassaying and
other cell biology techniques.
xv
I would like to thank all the friends I have made at the University of Pittsburgh for their
wonderful friendship.
At last, I would like to thank the most important people in my life: my father Li Xuelin,
my mother Shuai Yeying, my sister Li Hui and my husband Mao Wei. Thank you so much for
everything you have done for me. Without you, I could accomplish nothing.
xvi
1.0 BACKGROUND
Due to the potential applications in various areas and advantages over non-living sensors, cell-
based biosensors attract more and more attentions. However, current research on cell-based
biosensors is still in the early stage. The promise of cell-based sensing for drug discovery,
diagnostics, tissue engineering and pathogen detection is far from being realized [1]. In this
dissertation, we did two parts of work on the development of cell-based biosensors. First, we
developed acoustic wave sensor systems for cell biological studies. Second, we precisely
controlled cell shape using soft lithography techniques and investigated the effect of cell shape
on cellular function, which provides insight into the “design criteria” for engineering of a hybrid
interface that preserves biological responses of interest and is crucial to increasing the sensitivity
of cell-based biosensors. In this chapter, background information on these two topics will be
reviewed first.
1.1 SENSORS, BIOSENSORS AND CELL-BASED BIOSENSORS
A sensor is a device that responds to a stimulus and generates a signal that can be measured [2].
The stimulus could be physical, chemical or biological and the output signal is usually electrical
one that can be processed by electrical devices, such as an analog current or voltage, a stream of
digital voltage pulses, or an oscillatory voltage with amplitude, frequency and phase related to
1
the value of input quantity [2]. Sensors can be categorized into different classes, based on
different standards. For example, according to the applications or the physical quantities that can
be measured by sensors, they can be classifies as position sensors, force sensors, strain sensors,
pressure sensors, temperature sensors, chemical sensors, biosensors, etc. According to the type of
energy transfer they detect, they can be classified as mechanical sensors, optical sensors,
magnetic sensors, thermal sensors, radiant sensors, acoustic sensors, etc [3]. In this dissertation,
our research focus is on biosensors.
Biosensors (Figure 1) are those devices combining the biological sensing elements (such
as microorganisms, enzymes, antibodies, antigens, nuclear acids, cells, etc) with
physicochemical transducers (including optical, electrochemical, thermometric, piezoelectric or
magnetic transduction systems) to detect biological events/elements (e.g., cell behavior, enzymes,
antigens, and nuclear acids) [4]. In the past twenty years, biosensor research has grown rapidly
with the development of the biochemistry, molecular biology, immunochemistry, and the
development in fiber optics and microelectronics [5]. Biosensors have exhibited potential
applications in the areas of biomedicine, bacterial and viral analysis, and cellular and molecular
pharmacology. Various sensor technologies can be used to implement the biosensor platforms,
including microelectronics and electrical technology, optical technology, acoustic wave
technology, electrochemical technology, etc [3].
2
Electric signal
Biological events/elements
Molecular reorganization Layer (e.g. microorganisms, enzymes, antibodies, antigens, nuclear acids, cells etc)
Signal transducers
Figure 1. Principle of biosensors.
Cell-based biosensors are sensors combined with living cells. Compared with non-living
sensors, cell-based sensors offer a lot of potential advantages. For example, they are able to
detect unanticipated threats; they can relate the measurement data to cell pathology and
physiology; they enhance the stability of enzymes, receptors, or antibodies in biological systems.
The applications of cell-based biosensors are potentially extensive. First, by treating cells as the
research objectives, cell-based biosensors can be used in a lot of fundamental biological research.
Second, cells are the sensitive elements that are able to sense various biophysical or biochemical
signals and transduce them into changes in cell behaviors or properties that can be measured by
biosensor systems. So cell-based biosensors can be used for drug development (assess the
efficacy and toxicity of candidate drugs), medical diagnostics (predict clinical outcome), cell-
based therapy, detection of chemical and biological agents, etc [1]. Current research on cell-
based biosensors is still in the early stage. There are a lot of challenges on the development of
cell-based biosensors: (1) Cells are environmentally sensitive. Cells must be cultured in the
viable and sterile environment. (2) The inherent viability in cells is large. Even for the
genotypically identical organisms, they differ from each other. The variability could be induced
3
by differences in the cellular micro-environment, protein expression, receptor number and other
factors. Therefore, in order to obtain producible results, engineering the living/non-living
interface is crucial. (3) The relationship between input and output signals is hard to define.
Controlling over variability in the microenvironment of cells preserves biological responses of
interest and is crucial to increasing the sensitivity of cell-based biosensors. (4) Transduction of
cellular output to the signals that can be measured and analyzed is complex. Therefore,
development of innovative sensing technologies is crucial. Based on the above challenges, in this
dissertation, we applied acoustic wave devices for cell biological studies and precisely controlled
the adhesive islands’ shape using soft lithography techniques.
1.2 ACOUSTIC WAVE SENSORS
As its name implies, an acoustic wave sensor is based on the transduction mechanism between
electrical and acoustic energies [2]. In acoustic wave devices, the acoustic waves propagate
through or on the surface of the device. The velocity and amplitude of the acoustic wave can be
altered by the changes in the acoustic wave propagation pathway, including the mass changes
and viscoelasticity changes in materials bounded on the device surface, and force changes
applied on the device surface. Therefore, acoustic wave devices as sensors have been used to
measure gas or protein concentration, the thickness of thin film, liquid viscosity, polymer
viscoelasticity, etc [6, 7]. The acoustic waves can be generated by piezoelectric, magnetostrictive,
electrostrictive, photothermal, and other mechanisms, but most acoustic wave devices are based
on piezoelectric crystals, such as quartz crystal and ZnO [8]. In this section, first, a brief
overview of the fundamental background on acoustic wave devices will be given, including
4
piezoelectricity, acoustic wave in elastic media, device configurations and the measuring
techniques. Then, classification and comparison of acoustic wave sensors will be described.
Finally, the TSM and Love mode acoustic wave sensors that will be used in this study will be
reviewed.
1.2.1 General Background
Piezoelectricity
Most of the acoustic wave sensors are based on piezoelectric mechanism. The
piezoelectricity exists in the crystals that lack a center of inversion symmetry. The definition of
piezoelectricity is that the electrical charges on opposing surfaces of a solid material are
generated when torsion, pressure, bending, etc. are applied along an appropriate direction.
Conversely, the application of the polarizing electric field may cause mechanical deformations,
which is called the “converse piezoelectric effect”. The piezoelectric effect reflects the coupling
between the mechanical energy and the electric energy domain of the material.
Piezoelectric constitutive equations are developed to describe the electrical and
mechanical interactions in piezoelectric materials. Four sets of piezoelectric constitutive
equations are commonly used. In matrix notation, these equations can be written as follows [9]:
5
⎩⎨⎧
+−=−=
DhSEDhScTs
D
β
'
(1.1a)
⎩⎨⎧
+=+=
EdTDEdTsS
T
E
ε
'
(1.1b)
⎩ (1.1c) ⎨⎧
+−=−=
DgTEDgTsS
T
D
β
'
⎩⎨⎧
+=+=
EeSDEeScT
S
E
ε
'
(1.1d)
where the matrixes with superscript “’” are transpose of these matrixes, T is stress tensor, S is
strain tensor, D is electrical displacement tensor, and E is electric-field intensity tensor. In these
constitutive equations, there are twelve different constants. Their physical meanings are as
follows: and are the stiffness coefficient tensors under the condition of a constant electric
displacement and electric-field intensity, respectively; and are compliance coefficient
tensors under the condition of a constant electrical displacement and electric-field intensity,
respectively; is the piezoelectric charge coefficient tensor; d is the piezoelectric strain
coefficient tensor;
Dc Ec
Ds Es
h
g is the piezoelectric voltage coefficient tensor; e is the piezoelectric stress
coefficient tensor; and are permittivity tensors under the condition of a constant stress
and strain respectively; and and are permittivity tensors under the condition of a constant
stress and strain respectively.
Tε Sε
Tβ Sβ
Acoustic Wave in Elastic Media
A general approach to theoretically describe acoustic wave propagation is based on
solving a set of wave equations by applying boundary conditions between different layers. It is
necessary to describe the wave equations for non-piezo and piezoelectric materials here. Based
6
on the Newton’s second law of motion and the elastic constitutive equation, the wave equation
for a non-piezoelectric solid can be derived as follows [8]:
jmlikijkllikijkl uuuC..
,
.
, ρη =+ (1.2)
where and ijklC ijklη are the fourth-rank elastic coefficient tensor and the fourth-rank viscoelastic
coefficient tensor. For piezoelectric materials, the wave equation becomes
jmlikijkllikijklkiijk uuuCe..
,
.
,, ρηφ =++ (1.3)
where is the third-rank piezoelectric stress coefficient tensor. ijke
According to the directions of particle displacement and wave propagating, acoustic
waves are classified into shear wave and compressive wave. In a shear wave, the wave
propagation is normal to the particle displacement, while in the compressive wave, the particle
motion is in the same direction as the wave propagation [8].
Device Configurations
The configurations of acoustic wave devices include type and geometry of transducers
and the electronic circuitry that is connected to the transducers. According to the number of the
transducers, acoustic wave device can be classified into two main categories: one-port and two-
port devices [2].
One common example of the one-port device is the bulk resonator (Figure 2). Bulk
waves are propagating through the whole substrate. They are generally excited by the parallel
electrodes coated on both sides of the excited volume of a piezoelectric material. The input and
output ports are both served by the parallel electrodes on both sides. The surface acoustic wave
(SAW) resonators could be one-port device also. The details of the SAW resonators are given in
the next paragraph.
7
Figure 2. Structure of bulk acoustic wave resonator..
The examples of two-port devices are the SAW delay lines and resonators. In SAW
device, the acoustic wave propagation is confined to the surface. Usually, a surface acoustic
wave is excited on the piezoelectric crystal using lithographically patterned interdigital
transducers (IDTs) (Figure 3). The relative signal levels and phase delay between input and
output ports constitute two responses. When an alternative voltage is applied between alternately
connected electrodes, a periodic strain field is generated in the piezoelectric crystal underneath.
Thus, an acoustic wave perpendicular to the finger direction is excited by each finger pair.
Interference results in a maximum SAW magnitude if
vnf n λ12 +
= (1.4)
with the excitation frequency f, and the acoustic wave velocity v of the particular mode..
Therefore, the frequency is determined by the geometry of IDTs.
8
U
λ
a b
Figure 3. Interdigital transducer, formed by patterning electrodes on the surface of piezoelectric materials for the excitation of surface acoustic waves [2, 10]
The SAW resonators may be either one-port or two-port devices (Figure 4A), in which
one or two IDTs are enclosed between two reflectors [2, 10]. For the one-port resonator, its
properties are much like those of the bulk resonators. Delay lines (Figure 4B) are always two-
port devices having two IDTs deposited on the piezoelectric substrate with center to center
distance equal to a whole number of wavelengths. Compared with delay lines, the advantage of
resonators is that the insertion loss of the resonator is smaller, which decreases influence of the
phase noise on the measurement and results in an improved sensor resolution. However, in
contrast to delay lines, the phase characteristics of resonators are widely nonlinear and dependent
on damping. In addition, the sensor reaction to different external signals is very complex. Hence
it becomes difficult to separate the effects of different physical quantities. Another disadvantage
of typical resonators is that additional device area is needed for the reflective elements.
Therefore, in most cases, delay lines are used in sensing applications.
9
A B
Figure 4. Geometry of two-port SAW devices: Two-port SAW resonator (A) and delay lines (B) [10].
Measurement Techniques
Basically, there are two kinds of techniques for measurements of acoustic wave devices:
oscillator-based measurements and phase/amplitude-based measurements [8].
The oscillator circuit (Figure 5) has to be satisfied two conditions in order to maintain a
continuous oscillation. If we assume the impedance of acoustic device is Z, the loop gain
condition is as follows:
1)( ==⋅+
AARZ
Z β (1.5)
The loop phase condition for the oscillation is
πβ nA 2)arg( = (1.6)
In real circuits, the loop gain may be larger than unity. An amplifier with automatic gain control
(AGC) can adjust the loop gain to maintain a value of unity [11]. Moreover, changes in the
insertion loss can be monitored by means of the AGC control voltage [9].
10
Z
R
Vi Vo A
Figure 5. Oscillation circuit of bulk acoustic wave (BAW) and surface acoustic wave (SAW) devices.
In the SAW device, the dual delay line configuration is widely used. The signal of the
sensor is mixed with a reference one through a mixer, resulting in a low-frequency signal.
Therefore, common influences to both oscillators, such as temperature changes, are eliminated.
The low-pass filter is required to eliminate the sum of the frequencies and other higher harmonic
products after the mixer. Using this dual delay line configuration, the frequency shift of the
sensor can be detected with eliminating the influences of temperature, aging etc.
For the in-depth device characterization, a network analyzer (NWA) is an ideal tool.
Compared with an oscillator setup, the influence of an NWA itself on the measurement can be
ignored. A measurement of the scattering parameters Sij provides a complete characterization of
two-port devices such as resonators or delay lines (scattering parameters Sij are a set of
parameters describing an electrical two-port device; S parameters give a measure of the ratio of
the amplitudes of the incident and the reflected electromagnetic waves at both ports). The
forward-scattering parameter S21 is a measure of the ratio of the outward traveling wave at port 2
to the incident voltage at the port 1 (Figure 6A). The reflection coefficient S11 describes the ratio
of incident and reflected voltages at the port 1 (Figure 6B). In order to characterize a single IDT
or a one-port resonator, e.g., TSM resonator, S11 may be transformed into the impedance or
11
admittance domain. To avoid unwanted influences of reflected acoustic waves from neighboring
edges or transducers, those structures should be thoroughly covered with an absorbing material.
In addition, a one-port resonator can be measured by impedance analyzer, shown in Figure 6C.
A B C
Network analyzer
S12
AbsorberSAW Device
Network analyzer
S11
Impedance analyzer
One–port BAW Device
Figure 6. Devices characterized with network analyzer by measuring forward-scattering parameter S21 (A) the reflection coefficient S11 (B) and a one-port resonator is characterized by impedance analyzer (C) [10]
1.2.2 Classifications and Comparisons of Acoustic Wave Sensors
In the previous sections, we have discussed some general background on acoustic wave devices.
In this section, the classifications of acoustic wave sensors will be discussed and we will
compare the different acoustic wave sensors then. The acoustic wave sensors can be grouped into
several classes according to different mechanisms. For example, according to the geometric
shape of the sensor, one can differentiate between acoustic modes in a half-space (solid with one
free boundary), a plate (solid with two boundaries), layered structures (half-space with parallel
layers), and bulk modes (solid restricted in all dimensions), while according to the direction of
12
polarization of particle motion relative to the device surface, the acoustic wave sensors can be
categorized into sagittal polarization, shear-horizontal polarization and longitudinal polarization
sensors (Table 1).
Table 1. Classification of acoustic wave sensors [10]
sagittal
polarization shear-horizontal polarization
longitudinal
polarization
Half-space Rayleigh waves Leaky wave, surface skimming bulk
wave (SSBW)
QLL-SAW
Plate FPW(Lamb
wave) SH-APM
Layered
structures Love waves
Bulk mode TSM
The geometric shape of the sensor determines the sensitivity to mechanical interactions at
the device surface, i.e. mass loading or viscous coupling, which is of primary interest in sensor
applications. Generally, the sensitivity is related to the ratio of the square of surface particle
velocity to the average density of acoustic energy within the device. Therefore, Rayleigh waves
are extremely sensitive since the acoustic energy is essentially confined within a surface region
less than one wavelength thick. Furthermore, plate modes tend to become more sensitive with
decreasing plate thickness.
The polarization mode determines the possibility of acoustic wave sensors in liquid
applications. For example, the well-known Rayleigh waves with the sagittal polarization are not
suitable for liquid sensing in most cases since strong attenuation occurs due to the generation of
13
compression waves into the liquid because of the vertical component of particle motion. This
compression wave excitation can be avoided if the wave mode used is slow enough to allow only
evanescent waves in the adjacent liquid. Flexural plate mode (FPW) is an example of such waves.
Since liquids do not support shear waves, modes exhibiting shear-horizontal polarization can be
used in liquid sensing applications. In isotropic materials, there is no shear-polarized true surface
wave. However, in the case of a strongly anisotropic piezoelectric crystal, as well as layered
structures, surface-localized shear modes can exist, such as Leaky wave, surface skimming bulk
wave, Love waves, shear horizontal acoustic plate wave (SH-APM) and so on. Thickness shear
mode (TSM) wave, which is a shear-horizontal polarized bulk wave, is widely used in liquid and
gas sensing applications.
Table 2 lists selected properties of several acoustic sensors for liquid sensing applications.
As discussed, the sensitivity of the devices is related to the geometric shape of the sensor. For the
devices with a plate structure or bulk mode, such as Lamb mode, SH-APM and TSM devices, the
operating frequency and the sensitivity are relative low. The reason is that improvement of
frequency therefore increasing sensitivity demands reducing the membrane thickness, thereby
also reducing the mechanical robustness. For half-space structures, the sensitivity is only
determined by the working frequency. Although increasing frequency can be obtained by
decreasing the IDTs periods, it simultaneously demands more complex RF electronics. However,
layered structure, such as Love mode sensor, shows its advantage on the device sensitivity since
its sensitivity can be increased up to a certain limit by optimizing the waveguide thickness. This
does not affect the mechanical robustness.
Another important properties of devices are the cross sensitivities with respect to changes
of other physical quantities, such as pressure or temperature. High mechanical robustness of
14
devices can reduce pressure-sensitivity. Therefore, the devices with layered or half-space
structure have low pressure sensitivity. Temperature sensitivity can be reduced either by
choosing a proper crystal cut (TSM, Rayleigh, SSBW) or material system (Love mode). To
avoid electrode directly exposed to the liquids, in some devices, e.g. SH-APM, Lamb mode, the
electrodes are situated on the surface opposite the liquid contact. Another method is the device
surface is covered by an inert functional layer, such as Love wave.
Among these acoustic liquid sensors, TSM device is the simplest and the oldest. It has a
simple structure, relatively low fabrication cost and easy measurements due to the low operating
frequency. In addition, the theory of TSM device is relatively mature. Therefore, although TSM
device has some intrinsic limits such as relatively low sensitivities, it has been widely used as
sensors in gas and liquid phases. Love mode acoustic wave sensors have many advantages e.g.
high sensitivity, mechanical robustness, and the protection of the IDTs due to the waveguide.
Therefore, it is one of the most promising techniques used in liquid phase. However, there are
still a lot of unsolved problems for Love mode devices as sensors, such as theoretical modeling,
measurements etc. Therefore, in this study, we choose these two modes as cell-based biosensors
for cell biological studies.
Table 2. Selected properties of acoustic wave sensors [10]
Property Love TSM SH-APM Lamb SSBW,
leaky wave
Mode waveguide Bulk Reflected
plate plate Surface
Polarization Shear-hor. Shear-hor. Shear-hor. Sagittal Shear-hor.
Frequency (MHz) 50–500 2–20 50–200 1–10 50–500
15
Table 2 (continued)
Cross sensitivities
(Electrical
liquid
properties)
(temp.)
Pressure
(temp.)
electrical
liquid
properties)
Pressure,
electrical
liquid
properties
(temp.)
Pressure,
electrical
liquid
properties
(temp.)
Electrical
liquid
properties
(temp.)
Sensitivity
determined by
Waveguide
thickness
Frequency
Plate
thickness
Plate
thickness
Membrane
thickness Frequency
Frequency
determined by IDT period
Plate
thickness
Plate
thickness
IDT period
Membrane
thickness,
IDT period
IDT period
Fabrication cost 0 – to 0 + + –
–, none/small; 0, intermediate; +, large; ++, very large
1.2.3 Thickness Shear Mode (TSM) Devices
The thickness shear mode (TSM) devices (Figure 7A) commonly indicate a thin disk of the AT-
cut quartz plate with metal electrodes on both sides. The AT-cut quartz slice is prepared by
cutting a quartz crystal with an angle of 35.10 degrees to the optical z-axis (Figure 7B). Since
the AT-cut quartz crystals show a high temperature stability, this particular cut has become the
most suitable one for TSM resonators.
16
BA
d
Figure 7. Circular TSM quartz resonator (A). A quartz plate is cut with an angle of 35.10 degrees with respect to the optical z-axis. (B) [10].
Because of the piezoelectricity and the crystalline orientation of the AT-cut quartz plate,
the application of a voltage between the two electrodes produces a shear deformation in the plane
with the crystal surface (Figure 7A). When an alternative voltage is applied between the two
electrodes, shear waves of opposite polarity are generated at the either side of the crystal. The
resonant condition for the TSM resonator can be determined by tracing the path of one
propagating shear wave, which may be thought to start at the top electrode, propagate through
the thickness of the quartz thickness, reflect at the bottom surface with phase shift of π, and then
return to their origin where it reflects again with phase shift of π. Therefore, the total round-trip
phase shift is
πφ 22 += khr (1.7a)
svk //2 ωλπ == (1.7b)
where h is the thickness of the crystal, λ is the acoustic wave length, is the shear wave
velocity. When the total phase shift is an integer multiple of 2π, constructive interference of
sv
17
incident and return waves occurs and results in resonance. Therefore, the mechanical resonance
occurs when the thickness of TSM resonator is an odd multiple of half of the acoustic
wavelength.
)2
(λNh = (1.8)
Therefore, the resonant frequency is expressed as:
hNvf s
N 2=
(1.9)
TSM resonator can only excite the odd harmonics. At the resonance, a standing shear
wave is established with the maximum values of the displacement occurring at the crystal surface,
which makes the device sensitive to surface perturbations [2]. Using piezoelectric materials, the
mechanical resonances of devices are electrically excited or detected. Therefore, it is necessary
to look at the electrical characteristics of TSM resonators to describe the conversion between the
mechanical and electric parameters. So far, two types of equivalent-circuit models are used to
describe the resonators: the distributed (or transmission line) model and the lumped-element
model.
A
B
C0 jx
11:N
hq /2 hq/ 2
D
Zs F
Figure 8. Transmission line model to describe the near-resonant electrical characteristics of the resonator
18
A transmission line model (TLM) (Figure 8) is based on a one-dimensional
electromechanical model of acoustic wave generation and propagation in piezoelectric and non-
piezoelectric layers [12]. In this model, a piezoelectric material is described by a three-port
element, one electric port and two acoustic ports. In the case of the resonators with loading on
one side, one acoustic port is considered as a short circuit and the other acoustic port is loaded
with mechanical impedance Zs, representing the surface “loading condition”. Therefore, the
generalized input electric impedance of resonator sensors can be obtained as
⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜
⎝
⎛
−
−−=
α
α
αω cot1
2tan2
11 2
0
q
L
q
L
ZZj
ZZj
KCj
Z
(1.10)
The abbreviations in equation (1.10) are calculated with the following equations:
q
ce
Kε
22 =
(1.11a)
qqq ch /ρωα = (1.11b)
qqq cZ ρ= (1.11c)
qq h
AC ε=0
(1.11d)
where K2 is the electromechanical coupling coefficient of quartz, α is the acoustic phase shift in
the quartz crystal and Zcq is the characteristic acoustic impedance of the quartz crystal [12].
The electrical impedance can be separated into a parallel circuit consisting of the static
capacitance Co and motional impedance Zm. According to equation (1.10), the motional
impedance Zm can be written as follows:
Lmm
q
Lq
Lm ZZ
Z
ZjZZ
KCK
CjZ +=
×−
+⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜
⎝
⎛
−= 02
0
2
0
2tan2
1
14
11
2tan2
1
α
αωα
α
ω
(1.12)
19
Zm0 and Zm
L represent the unloaded quartz and the load respectively.
By taking the approximation of))/((4
2tan 22 απα
α−≈ N
, the unloaded quartz near the resonant
frequency is expressed as
qqq
m RLjCj
Z ++= ωω10
(1.13)
For small loads ( 2tan2 α
<<q
l
ZZ
), ZmL simplifies to
q
LLm Z
ZKC
Z 20 4
1 αω
= (1.14)
Therefore, for resonators with small loadings, near the resonance frequency, TLM can be
transformed into the equivalent circuit with lumped elements, which is called Butterworth-Van-
Dyke circuit (Figure 9).
C0*
C0*
Cm Lm Rm ZmL Cm Lm Rm A B
Figure 9. Butterworth-van-Dyke (BVD) equivalent circuit of unperturbed TSM resonator near the series resonant
frequency (A); Modified Butterworth-van-Dyke equivalent circuit of TSM resonators with small loadings (B)
In case of the resonators as practical sensors, the parasitic capacitance not only includes
the static capacitance C0 in equation (1.11d), but also includes a capacitance Cp that arises in the
test fixture. In the BVD circuits, we use C0* to represent the parasitic capacitance. The resonance
frequency is a critical parameter for TSM resonators in sensing applications. According to BVD
circuits, the quartz resonator shows two different resonance frequencies [2]. The series resonant
20
frequency is defined as the frequency at which the motional reactance is zero. According to
the above criterion, for the unperturbed resonator, the series resonance frequency is given by
sf
2/1)(21
mms CL
fπ
= (1.15)
The parallel resonance frequency is defined as the frequency at which the total
reactance is zero.
pf
2/1
0
)*
11(121
⎥⎦
⎤⎢⎣
⎡+=
CCLf
mmp π (2.16)
In the measurement, the series resonance frequency is obtained by measuring maximum
value of the in-phase electrical admittance of the resonator Gmax, while the parallel resonance
frequency is the characteristic frequency obtained by using the active oscillator. Equation (1.10)
shows the relationship between the input electrical impedance of the resonator and the surface
mechanical impedance of the loadings. Therefore, the input electrical impedance provides
information about a variety of different material-specific properties, such as mass of thin films,
viscoelasticity of liquids or shear moduli of viscoelastic materials. Different surface
perturbations have different the surface mechanical impedance (Zs) (Table 3). According to
expressions of the mechanical impedance for different loadings, the mechanical properties of the
loading materials can be extracted from the input electrical impedance of the resonators near the
resonance.
21
Table 3 Surface mechanical impedance (Zs) for different surface perturbations
Surface perturbation Surface mechanical impedance (Zs)
Unperturbed 0
Ideal mass layer sjωρ
semi-infinite Newtonian liquid )1()2/( 2/1 jLL +ηωρ
Rigid mass + semi-infinite
Newtonian liquid )1()2/( 2/1 jj LLs ++ ηωρωρ
semi-infinite viscoelastic layer 2/1)( fGρ
Finite viscoelastic layer )tanh()( 2/1fl hG γρ
In recent years, TSM resonator has became a versatile tool in bioanalysis. Monitoring the
adsorption of proteins is the primary application area for TSM resonators. The proteins could be
physically adsorbed onto the surface to form a monolayer, or protein multilayer deposited by the
Langmuir-Blodgett techniques or self-organization processes [13-15]. Immunosensors are
another important application for TSM resonators [16-19]. More recently, TSM resonators
combined with cells have been used to monitor cell behaviors. Several types of cells were
investigated in these studies, including endothelial cell, osteoblasts, MDCKI and II cells, 3T3
cells, VERO cells and CHO cells [20-23]. In most of studies, the dynamic processes of cell
attachment and spreading on the resonator surface have been monitored. The results have
demonstrated that resonant frequency shifts and motional resistance changes are related to the
processes of cell attachment and spreading. However, the frequency shift due to cell monolayer
attachment is significantly smaller than expected by the Sauerbrey equation for cell mass.
Therefore, Redepenning et al. pointed out that cell layer attaching to the resonator cannot be
22
treated as a rigid mass [24]. Later, studies by Jansoff et al. and Fredriksson et al. showed that the
present of a confluent cell layer attaching to the resonator increased both energy dissipation and
energy storage on the resonator surface, which indicates that the cell has to be treated as a
viscoelastic material [20, 25]. Recent studies by Wegener et al. demonstrated that the signals of
TSM resonators are contributed to many factors, such as cell-substrate interactions, viscoelastic
properties of cells, and the properties of the interfacial layer between cells and substrate [23].
Although previous studies have shown acoustic wave sensors can detect various changes in the
acoustic wave propagation pathway, current research on acoustic wave devices as cell-based
sensors is only focused on monitoring cell attachment and spreading. There is no theoretical
model to describe the relationship between electrical measurements and cell behaviors as well as
properties. No studies have been performed to characterize cell viscoelasticity by using TSM
resonator sensors. Therefore, in this study, we developed a theoretical model to predict how
electrical measurements of TSM resonators change with the cell layer and interfacial layer
between the cell layer and sensor surface and studied the changes in cell-substrate distance and
cell viscoelasticity during cell attachment and spreading, and formation of focal adhesion, as
well as under cytoskeleton disruption. Using our theoretical model, we also characterized the
viscoelasticity of cell monolayer by TSM resonators.
1.2.4 Love-mode Acoustic Wave Sensors
Since liquids do not support shear waves, modes can be used for liquid sensing if they
exhibit shear-horizontal polarization. In isotropic materials, there is no true shear-polarized
surface wave. However, for a strongly anisotropic piezoelectric crystal and layered structures,
surface-localized shear modes can be obtained. In some anisotropic materials, a surface
23
skimming bulk wave (SSBW) occurs (Figure 10A) [10]. To satisfy the stress-free boundary
condition, the SSBW is propagating into the bulk materials. Another mode supporting shear-
horizontal waves is leaky SAW(Figure 10B) [10]. The wavevector and the direction of energy
flow are not parallel to each other. Lithium tantalite (LiTaO3) with the Y-rotated cut along the X-
axis support the leaky SAWs.
For the surface skimming bulk wave and leaky SAWs, the acoustic waves are
propagating deeper (1-5 wavelength) into the bulk material, which reduces the sensitivity to the
surface perturbation. One approach to increase the sensitivity is to apply a dielectric layer onto
the top of the propagating surface to trap the acoustic energy near the surface (Figure 11). This
mode is called wave-guided SH-SAW or Love mode. Any materials that support shear wave
shear waves such as SSBW, SH-SAW and leaky waves can be used as a substrate of Love mode
devices. For example, as a standard material, Y-rotated quartz (ST-quartz) which supports
SSBW has often been used. Furthermore, 36o YX-LiTaO3 that supports leaky SAWs is
advantageous because of its very high electromechaical coupling coefficient and moderate
temperature dependence. Amorphous silicon dioxide (SiO2) and also polymers have been
reported as overlay materials.
A B
Figure 10. Two surface-localized shear modes in anisotropic materials (SSBW (A) and Leaky SAW (B)) [10]
24
IDTs
Wave-guided layer
Figure 11. Love wave sensor. [10]
Although Love mode acoustic wave devices have a lot of advantages in terms of using in
the liquid phase, current research on Love mode acoustic wave devices as biosensors is very
limited due to complexity of theoretical modeling and experimental measurements. Theoretical
modeling of Love mode devices is important to optimize the sensor performance and predict the
electrical signals changes due to the perturbation on the sensor surface. The general method is
based on seeking the solution of the wave equations that satisfies the boundary conditions
between each layer. This method can accurately describe the acoustic waves. However, it is
computationally intensive and difficult to relate physical changes to corresponding numerical
changes in the model, especially in the case of considering the loss of materials. Therefore, this
analytic method is usually used to predict the electrical signals changes induced by the mass
changes in the loading layer. Perturbation theory has also been used in the theoretical modeling
of Love mode devices to evaluate the effects of small system variations. Basically, the wave
equation without perturbation is solved first and then the solutions are adapted to perturbation
equations to solve the effect of small changes in the sensor system [3, 28, 29]. The assumption in
perturbation theory is that the changes in the system parameters are very small. The perturbed
25
termed are often replaced by the exact, unperturbed solution and only the quantities of interest
are unknown in the equations. Therefore, the perturbation method is limited on the modeling of
the small changes of physical parameters in a system. Transmission line model is one method
that has been widely used for the modeling of acoustic wave devices. An analogy between
mechanical structures and transmission lines is used to describe the properties of propagation and
attenuation along the acoustic devices in this method [30]. Recently, the transmission line model
has been applied for the modeling of Love mode acoustic wave devices as sensors [31]. It has
been demonstrated that this model provides an efficient way to describe the behavior of the
sensor loaded by viscoelastic masses in a viscous liquid. However, this method is complex for
modeling the complex loading on the device surface, such as the multilayer loadings in cell-
based biosensors. Therefore, it is necessary to develop a simple and precise method to describe
how the two characteristics of the Love mode acoustic wave device (e.g. phase velocity and
insertion loss) change with multilayer loadings. The loading layers could be pure elastic,
viscoelastic or viscous layers. In this dissertation, transfer matrix method will be used in
modeling of Love mode devices as cell-based biosensors.
1.3 MICROFABRICATION AND SOFT LITHGRPAHY
Micro-fabrication or micromachining are the techniques for fabricating of miniature structures, at
micron or smaller scales. It was originally used for semiconductor devices in integrated circuit
fabrication. Now they are extended to be used for the fabrication of sensors, microreactors,
microelectromehcanical systems (MEMS), microanalytical systems and micro-optical systems.
The basic techniques include deposition of thin film, patterning, etching and so on. The thin film
26
deposition processes include the simple spin-on method, where a material in liquid form is
deposited on the wafer and the wafer is spun at high speeds to evenly spread a thin film of
material across the substrate, and more complex physical and chemical methods that utilize
complex chemical reactions and physical phenomena to deposit films. Patterning techniques are
used to create features on the micrometer or nanometer scales in the thin film layers. The most
powerful of patterning techniques is the photolithographic (also optical lithography) process. It
uses light to transfer a geometric pattern through a photomask to a light-sensitive chemical
(photoresist) on the substrate. Usually, photolithography involves three primary steps: deposition
of a photosensitive emulsion layer, ultraviolet exposure of photoresist to some pattern and
subsequent develop of the wafer to remove unwanted photoresist and etching of materials in
desired areas by physical or chemical means. The patterned photoresist serves as a material
template for the definition of microstructures. Patterns can be etched directly into the silicon
substrate, or into a thin film which in turn can be used as a mask for subsequent etches. Many
etching forms including wet-chemical, dry-chemical, and physical etching techniques are used in
MEMS processing.
Although photolithography is a dominant technology, it has some limitations in several
applications. For example, it is poorly suited for patterning non-planar surfaces; it is relatively
limited for the fabrication based on the materials other than photoresists; if working with other
materials, it is necessary to attach photosensitizer, which is not convenient [32]. In 1990’s,
another type of patterning technique called “soft lithography” was developed by the George
Whitesides Research Group at Harvard University [33]. Soft lithography uses soft, transparent
and elastomeric polydimethylsiloxane (PDMS) "stamps" with patterned relief on the surface to
generate features. The dimension of features generated by soft lithography is as small as 30nm.
27
Usually, the stamps can be prepared by molding prepolymers against masters by conventional
lithographic techniques. Up to now, a lot of techniques have been developed, including micro-
contact printing (μCP), replica molding [34], micro-transfer molding (uTM), micro-molding in
capillaries(MIMIC), solvent-assisted micro-molding (SAMIM), and so on. In the following, I
will briefly review several soft lithography techniques and their applications [32, 33].
Micro-contact printing technique uses the relief pattern on the surface of a PDMS stamp
to form patterns of self-assembled monolayer (SAMs) on the surfaces of substrate by contact.
For example, hexadecanethiol self-assembled monolayer (SAMs) can be patterned on the surface
of gold by micro-contact printing method (Figure 12). The patterned self-assembled monolayer
can act not only as a resist against etching, but also as a template in selective deposition.
Therefore, the materials that can be patterned using microcontact printing method are not limited
to SAMs. A variety of materials can be patterned, including liquid prepolymers, conducting
polymers, inorganic salts, metals, ceramics and proteins, by patterning SAMs at the molecular
scale followed with the deposition of other materials on the patterned SAMs.
Figure 12 Schematic procedures for microcontact printing of hexadecanethiol on the surface of gold [32]
28
Other types of soft lithography methods that are widely used are micro-molding related
techniques, including replica molding (REM) [34], microtransfer molding (μTM), micro-
molding in capillaries (MIMIC) and solvent-assisted micro-molding (SAMIM) (Figure 13). In
replica molding, PDMS stamp is molded against a photolithographic patterned master.
Polyurethane is then molded against the secondary PDMS master. In this way, multiple copies
can be made without damaging the original master. The technique can replicate features as small
as 30 nm [32]. In microtransfer molding (μTM), a PDMS stamp is filled with a prepolymer or
ceramic precursor and placed on a substrate. The material is cured and the stamp is removed. The
technique generates features as small as 250 nm and is able to generate multilayer systems [35].
In micro-molding in capillaries (MIMIC), continuous channels are formed when a PDMS stamp
is brought into conformal contact with a solid substrate. Capillary action fills the channels with a
polymer precursor. The polymer is cured and the stamp is removed. MIMIC is able to generate
features down to 1 µm in size [36]. In Solvent-assisted Microcontact Molding (SAMIM), a small
amount of solvent is spread on a patterned PDMS stamp and the stamp is placed on a polymer,
such as photoresist. The solvent swells the polymer and causes it to expand to fill the surface
relief of the stamp.
Soft lithography techniques have a lot of advantages over conventional patterning
methods. First, this technique is low cost, easy to learn and accessible to a wide range of users. It
has capacity of generating features on non-planar substrates. It can be used to pattern a wide
variety of materials, such as metals and polymers [37, 38]. In addition, it can produce the
patterns and structures with lateral dimensions from 30nm to 500μm, which extends the existing
nanolithographic techniques. However, there are several unresolved problems such as distortions
in the mask which induces the defects in patterns. Soft lithographic techniques have wide
29
potentials in fabrication of a variety of functional components and devices in the areas of optics,
microanalysis, display, MEMS, microelectronics, and cell biology. Especially, in cell biological
studies, soft lithography is well suited for generating micro-fluidic channels in PDMS, which is
useful for the study of cells. The patterning of bacteria and cells on the surface by soft
lithography provides new opportunities for sensing and detecting bimolecular as well as whole
cells and studying cell-cell interaction and interactions between cells and their surroundings. For
example, soft lithography can be used to immobilize the single-strained DNA on sensor surface
(such as microcantilever) to detect the DNA hybridization. Using soft lithography, many studies
have been performed to investigate the effects of cell spreading area and cell shape on cell
function. Previous studies showed that the degree of cell spreading influences DNA synthesis
[39], proliferation and apoptosis [40, 41], and focal adhesion assembly [42] in human and bovine
capillary endothelial cells; differentiation in mesenchymal stem cells [43]; and gene expression
and protein synthesis in primary rat bone cells [44].For a fixed spreading area, studies on cell
shape regulation are mainly limited to cell motility, i.e., lamellipodia extension and direction of
cell migration. For example, cells preferentially extended lamellipodia from their corners
regardless of their geometric shapes [45, 46]. During migration, narrow teardrop-shaped or
triangular cells moved predominantly toward the blunt end, while wide cells did not; rectangular
cells migrated outward towards their two ends without bias for either one, but circular or square
cells moved in random directions [33]. In this dissertation, using the soft lithography techniques,
we studied the effects of the aspect ratio of cell shape on collagen type I synthesis in human
tendon fibroblasts, examined the distribution of actin cytoskeleton, focal adhesions, and cell
traction forces in human tendon fibroblasts of various shapes and correlated them to the
expression of collagen type I in these tendon cells.
30
Figure 13 Schematic illustration of procedures for (a) replica molding [34], (b) microtransfer molding (μTM), (c) micromolding in capillaries (MIMIC), and (d ) solvent-assisted micromolding (SAMIM)[32].
31
1.4 CELL BIOLOGICAL BACKGROUND
In this dissertation, we applied acoustic wave sensing and soft lithography technologies for cell
biological studies. In this section, some background information on cell biology based on the
scope of this dissertation will be introduced. The section will begin with a discussion of cellular
architecture, especially for cell membrane and cytoskeleton that are associated with cell
adhesions and viscoelastic properties. Extracelluar matrix (ECM), cell-ECM adhesion, cell
traction force and cell viscoelastic properties will be explained then.
1.4.1 Cellular Structure
Plasma Membrane
The plasma membrane (Figure 14) consists of phospholipids and proteins. The
phospholipids with a phosphate group attached to one end are arranged in two layers. The lipid
part of the molecule is hydrophobic while the end with phosphate group is hydrophilic. In each
layer, the hydrophobic part of the molecule faces toward the other layer, and the hydrophilic end
of each molecule faces away from the membrane. The lipid bilayer is a basic structure of the
membrane and prevents most water-soluble material from entering cells. This phospholipids
bilayer is generally about 7.5 nm thick. Proteins in the membrane have hydrophobic and
hydrophilic regions and orient themselves in the lipid layer accordingly. Hydrophobic parts of
the protein associate with the interior of the membrane, while hydrophilic parts protrude to the
outside of the membrane. These membrane proteins can be classified into two different types: the
integral proteins that pass through the cell membrane and the peripheral proteins which stud the
inside and outside of the membrane. On average, proteins constitute approximately 50% of the
32
mass of the cellular membrane and serve approximately seven different cellular functions: some
proteins provide structural integrity to the cell membrane; some proteins serve as pumps to
transport ions across the cell membrane; some proteins as carriers transport substances within the
cell; integral proteins that form ion channels; receptor proteins that bind neurotransmitters and
hormones; enzyme proteins that catalyze reactions at the membrane surface; and glycoproteins
that are responsible for antibody processing and cellular adhesion to surfaces.
Figure 14 Cell plasma membrane consisting of phospholipids and proteins [47].
Cytoskeleton
Cytoskeleton is an internal three-dimensional network throughout the intracellular space.
It helps establish and maintain cell shape, cell movement and cell division. In addition, the
cytoskeleton allows positioning and moving organelles within the cytoplasm. It has been
estimated that up to 80% of the proteins of the cytoplasm and 20-40% of the water in the
cytoplasm are associated with the cytoskeleton [48]. Based on their diameter and their chemical
33
properties, the structural elements in the cytoskeleton are classified into three types: actin
filaments, microtubules, and intermediate filaments (Figure 15).
(a) (b)
(c)
Figure 15. Three types of cytoskeleton: (a) actin filaments, (b) microtubule and (c) intermediate filaments
Actin filaments are the smallest major cytoskeleton components, whose diameter is about
7nm. Actin filaments are polymers of protein actin. Normally, individual actin filaments
organize themselves into actin bundles and networks. The bundles consist of closely packed
arrays of parallel actin filaments and generally protrude from the cell surface, while the networks
consist of loosely packed, crisscrossed actin filaments and are present in the interior cells. The
actin bundle projections create fingers of membrane: microvilli and filopodia. The microvilli are
responsible for transportation of nutrients from the surrounding media into the cell or tissue,
34
while filopodia are responsible for cell attachment.Both actin bundles and networks are
connected to the plasma membrane via membrane-actin filament binding proteins [47].
Microtubules are the largest structures in the cytoskeleton and their diameter is approximately 25
nm. In contrast to actin filaments, these long microtubules never form two- or three-dimensional
networks and have an inherent polarity. Their negatively charged ends make the microtubules
radiate out from the nucleus of the cell while the positively charged ends elongate and make
contact with the actin network associated with the plasma membrane. This allows the
communication, transport and control from the center of the cell to the periphery. Microtubules
are responsible for chromosomes movement, polarity and overall shape of cells, the spatial
disposition of organelles, distribution of actin filaments and intermediate filaments etc.
Intermediate filaments are 10 nm in diameters, smaller than the 25 nm microtubules but larger
than the 8 nm microfilaments. The intermediate filaments are the most stable elements in
cytoskeleton and are though as a scaffold to support the entire cytoskeleton framework. In
addition, they are prominent in cells that withstand mechanical stress and are the most insoluble
part of the cell [48].
35
Figure 16. (A) Tensegrity model. (B) Force balance between tensed microfilaments (MFs), intermediate filaments (IFs), compressed microtubules (MTs) and the ECM in a region of a cellular tensegrity array. [49].
36
Previous studies show that cytoskeleton is not a passive gel. Cells generate tensile forces
through actomypsin filament in their cytoskeleton. People used “tensegrity” model to predict the
complex mechanical behaviors of mammalian cells (Figure 16) [49]. In the cellular tensegrity
mode, the whole cell is described to be a prestressed tensegrity structure. Tensional forces are
borned by actin microfilaments and intermediate filaments, and theses forces are resisted by the
compression by internal microtubule and extracellular matrix (ECM).. Therefore, when cells are
subjected to the mechanical forces or changes in mechanical properties of ECM (e.g. stiffness of
ECM), the force balance between cells and ECM are altered. Cells sense the external mechanical
signals at the adhesions between cells and ECM and change their cytoskeleton network structure
and tensile forces in order to obtain new force balance with underlined ECM. For example, when
large deforming forces are applied to the integrins, cytoskeletal filaments and linked intranuclear
structure can be found to realign along the applied tension force direction [50, 51]. Changes in
cytoskeleton structure and mechanics cause the global distortion of the whole cell shape. In the
above example, cells will change to the elongated shape along the applied field line. As talked
earlier, up to 80% of the proteins of the cytoplasm and 20-40% of the water in the cytoplasm are
associated with the cytoskeleton, including a lot of enzymes and substrate that mediate cellular
metabolism (e.g., protein synthesis, glycolysis, RNA processing, DNA replication) [52]. The
changes in cytoskeleton structure and mechanics induce the stress and shape change in the
molecules that are immobilized to the cytoskeletal scaffold. These shape changes of molecules
alter the biophysical properties (thermodynamics, kinetics), and hence biochemistry (e.g.
chemical reaction rates) will be altered [53]. Finally, these biochemistry changes will produce
various effects on cell behaviors (e.g. proliferation, apoptosis, ECM production). So, because of
this integrated structure, cytoskeleton governs how cells process and integrate locally-induced
37
signals. Through the tension-dependent changes in cytoskeleton structure and mechanics, cells
respond globally [54]. Because of the specific structure and properties, cytoskeleton plays a
critical role in regulating cell-ECM adhesion, determining cell mechanical properties and cell
mechanotransduction.
1.4.2 Extracellular matrix (ECM)
In vivo, the cellular membrane is surrounded by the extracellular matrix (ECM), a collection of
fibrous collagen proteins, proteoglycans (complexes of polysaccharides and proteins), and
glycoproteins (proteins linked to chains of carbohydrates), and adhesive glycoproteins
(fibronectin, laminin, vitronectin and tenascin). It can be found in the extracelluar environment
of all tissues and organs. Its compositions are changed with different cell type, tissue type and
culture conditions [55].
The ECM serves various functions [55, 56]. First, it provides cells the physical
microenvironment. It allows cell attachment and serves as a scaffold, guides cell migration
during embryonic development and wound repair. Second, the ECM contains many signaling
molecules that regulate various cellular functions, such as proliferation, differentiation and death.
In addition, previous studies have revealed that the molecular compositions of the ECM, the
geometric presentation of adhesive proteins in ECM, as well as its topography and mechanical
properties are important in cellular function [57]. Recent studies have demonstrated that the
cytoskeleton organization and focal adhesions on soft substrates are different from those on hard
substrates, which suggests that the cell probes the stiffness of the ECM and alters its structure
and function accordingly [56, 58]. On the other hand, these features of the ECM are determined
both by the cells that produce it and by the cells that grow on or inside the matrix.
38
1.4.3 Cell-ECM adhesion
In order to survive and grow, cells must attach to and spread on a substrate. When a cell contact
with a solid surface, adhesions are formed with ECM at specific sites. These adhesions are
mediated by integrins, which are transmembrane proteins with extracellular domains that bind
with low affinity to fibronectin (or vitronectin, or some other component of the ECM) and
intracellular domains that link to the cytoskeleton (Figure 17) [58, 59]. Before the activation,
integrins are freely diffusive within the cell membrane. However, after they bind to the ECM, the
integrins undergo a conformational change and cytoplasmic proteins (talin, vinculin, α-actin) are
recruited to the adhesion site. Those cytoplasmic proteins connect the integrins to the
cytoskeleton, or biochemically initiate or regulate intracellular signaling pathways. With physical
clustering of multiple integrins and recruitment of more cytoplasmic proteins, the size, adhesion
strength, and biochemical signaling activity of adhesions increase. These larger, clustered
structures of integrins and cytoplasmic proteins are commonly called focal adhesions (Figure
18). Focal adhesions are flat, elongated structures with several square microns in area and
separated from the substrate by 1 to 15 nm. Normally, these focal adhesions are located near the
periphery of cells. Development of focal adhesions is stimulated by the small GTPase Rho-A,
and is driven by actomyosin contractility [58].
39
Figure 17. Focal contact or adhesion [47]
Figure 18. Immunofluorescence image of the focal adhesions [56].
Via the focal adhesions, cell-ECM adhesion provides a common pathway for the signals
transmission. Through cell-ECM adhesion, ECM scaffold provides the mechanical support to
balance the tension in adhering cells. So, mechanical properties, topology of ECM, and
availability of adhesive proteins in ECM are critical determinations of cytoskeletal organization
and cell morphology, which govern a lot of cell behaviors. Conversely, changes in cytoskeleton
structure and mechanics alter the states of cell adhesion to ECM.
40
When cells contact to a solid surface, cells experience three stages: attachment, spreading
and formation of focal adhesions and stress fibers (Figure 19). The adhesive strength increases
with each stage. The stage of attachment involves integrin–ECM binding. During this state, the
binding between cells and ECM is very weak. With the binding of ECM components to integrins,
integrin clustering is induced and the affinity for the ligand increases. After that, the contact area
with substrates increases and actin microfilaments are developed. This state is considered as the
intermediate state between weak adhesion and strong adhesion. During the third stage, cells
organize their cytoskeleton and form actin-containing stress fibers. At the same time, the
integrins cluster together and cytoplasmic proteins are recruited to the adhesion site to connect
integrins to cytoskeleton to form focal adhesion. Focal adhesions enhance adhesion, functioning
as structural links between the cytoskeleton and the extracellular matrix and triggering signaling
pathways that direct cell function. After the formation of focal adhesion and stress fibers, a
strong and steady adhesion between cells and ECM substrate is obtained [60]. Cells modulate the
states of cell adhesion in many pathological and physiological processes, such as tissue
remodeling during morphogenesis and wound healing, cellular metaplasia, cell proliferation,
tumor cell matastasis, cytokinesis and cell apoptosis during tissue remodeling [60]. For example,
the weak adhesive state is consistent with cells undergoing apoptosis during tissue remolding, or
those undergoing cytokinesis. The intermediate state is characteristics of cells involved in the
response to injury during wound healing or in tissue remodeling during morphogenesis. The
strong adhesion shows cells are differentiated and quiescent.
41
weak adhesion intermediate adhesion strong adhesionweak adhesion intermediate adhesion strong adhesion
focal adhesion
integrin cytoskeletonECM
focal adhesion
integrin cytoskeletonECM
Figure 19. Three states of cell-ECM adhesion [60].
The existing methods for studying cell adhesion are very often based on the observation of
cell morphology using time lapse videomicroscopy [25]. The limitations of this method are time
consuming, labor intensive and qualitative in measurements. To overcome these limitations,
methods for quantitatively monitoring cell adhesion in real time are highly desirable. Recently, a
technique called electric cell–substrate impedance sensing (ECIS) was developed (Figure 20)
[61, 62]. This technique is based on the electric impedance measurement between one large
electrode and one small electrode. Cells are cultured on the small electrode and the large
electrode is served as a reference. The cells block the current flow path and change the
impedance, which is dependent on the cell-covered area, cell-substrate distance and electrical
impedance of cell layers. So this method provides quantitative information on cell morphology
and cell-substrate separation in real time. It can be used to measure cell attachment, motility, and
membrane capacitance. The impedance measurement provide the electrical properties of cells,
however, it cannot be used to measure cell mechanical properties.
42
Figure 20. Principle of electric cell-substrate impedance sensing (ECIS) for monitoring cell adhesion [63]
1.4.4 Cell viscoelasticity
Cells have been shown to exhibit a viscoelastic behavior in previous studies [64, 65]. When
deformed, the cell stores mechanical energy, which permits subsequent recovery of its shape. It
also dissipates mechanical energy through mechanical friction. The responses of the cell to
mechanical forces depend on the rate of mechanical loadings. The structural origin of cell
elasticity is widely believed to originate in the network of actin fibers within the cytoskeleton.
However, the origin of energy dissipation is less clear but is often thought to arise from viscous
mechanisms associated with the cytoplasmic fluids and membranes. As discussed earlier,
cytoskeleton plays a critical role in regulating the mechanical stresses and resulting deformations,
43
which has been recognized to regulate various cell function, such as proliferation, differentiation,
wound healing, protein and DNA synthesis, apoptosis, and cell motility [66]. The measurements
of viscoelastic parameters of cells are expected to give insight into the structure of the cortical
and internal cytoskeleton and are important to understanding cell function in many physiological
and pathological processes. It has been observed that cell motility is related to viscoelastic
change of the lamellipodium, which is explained as the polymerization of actin filaments [67].
Stretch can modify cell viscoelastic properties because it alters the balance of forces at the cell-
cell and cell-matrix adhesions and induces a change in cytoskeletal prestress [68]. In addition,
the changes in viscoelastic properties of living cells can be induced by the onset and progression
of diseases or surrounding biological environments. For hepatocellular carcinoma cells (HCC),
its viscoelastic properties probably have significance on tumor cell invasion and metastasis, in
which it determines the flow behavior of tumor cells in the circulation and thus whether such
cells can be arrested to form metastases [69]. Therefore, measurements of viscoelastic properties
of cells help understand the developmental processes underlying disease progression and also
facilitate the use of living cells as sensory elements in the biosensor systems for numerous
emerging applications such as drug screening and environment monitoring, etc. Current common
techniques for measuring cell viscoelasticity include micropipette aspiration, optical tweezer,
microplate, atomic force microscopy (AFM), magnetic bead rheometry and micro/nanoparticle
(Figure 21) [70]. Here I will briefly review these techniques.
44
Nucleus
AFM indentation
Micropipette
Micropipette Aspiration
Optical Trap
Optical Tweezers
Magnetic Twisting Cytometry
Microbead
Microplate method
Microplate
Nucleus
Micro/nano particle tracking
AFM indentation
Micropipette Optical TrapOptical TrapOptical Trap
Optical TweezersMicropipette Aspiration Optical Tweezers
Magnetic Twisting Cytometry
Microbead
Magnetic Twisting Cytometry
MicrobeadMicrobead MicroplateMicroplateMicroplate
Micro/nano particle trackingMicroplate method Microplate method
Figure 21. Schematic depiction of common experimental techniques for measuring cell visoelasticity [70]
Micropipette aspiration is one of the most widely used techniques for characterization of
cell viscoelasticity. During the measurements, the cell is aspirated into a glass pipette with an
inner diameter smaller than the cell. The cell deformation is measured by optical microscopy
[71]. Fine force resolutions are controlled by the applied pressure and the inner diameter of the
pipette. The force applied on the cell and cell deformations are used to exact cell viscoelasticity.
However, this technique requires the cell to undergo large deformations, which makes it
incompatible with some cell types.
Optical or laser tweezers techniques are used to manipulate functional beads attached on
cells. The laser refracts when it passes through a bead. According to the conservation of
momentum, the changes in photon momentum cause the changes in the bead’s momentum[72].
As such, the force is applied on the beads through the focal points of the laser. This force is
proportional to the perpendicular distance from the optical axis of the trap. The displacement of
the beads can be measured by tracking the refracted beam with photodiodes. This method is used
to measure the global viscoelastic properties of the single cell.
45
Microplate method is another widely used method for characterization of cell
viscoelasticity. Two microplates, one rigid, the other flexible are used in this method. The cell is
adhered on these two microplates. In the measurements, the rigid microplate is moved by
micromanipulator to deform the cells [73]. Compression, traction or oscillatory deformations can
be obtained by controlling the movement of the rigid microplate. The cell deformation can be
monitored by microscopy. The flexible microplate serves as a force sensor. Since the stiffness of
the flexible microplate is known by calibration prior to the measurements, the force applied on
the cell can be obtained by measuring the deformation of the flexible microplate under
microscopy.
As discussed, micropipette aspiration, optical tweezers and microplate method are
invasive methods for characterization of global cell viscoelasticity. In these methods, usually, the
creep response and stress relaxation are measured. The simple viscoelastic models with finite
springs and dashpots are used to explain the experimental results. However, the spring-dashpot
model is over-simplified, which cannot precisely describe the complex mechanical properties of
cells and makes the results non-reliable. More recently, dynamical viscoelastic properties of cells
were characterized. In these studies, the complex shear modulus (G*(ω)) from oscillatory
measurements over a wide frequency range were measured. G*(ω) is defined as the complex
ratio in the frequency domain between the applied stress and the resulting strain. The real and
imaginary parts of G*(ω) account for the elastic energy stored and the frictional energy
dissipated within the cell at different oscillatory frequencies.
AFM is one of methods to probe the dynamical viscoelastic properties of cells [74]. In the
measurements, the AFM probe is indented into and deforms the cell body. The cantilever
deflection is detected by the photodiodes. However, operation in aqueous environments, which is
46
often required to maintain normal cell function, may be complicated due to fluid-probe
interaction and reflection or refraction of the laser used for sensing [70]. Magnetic bead micro-
rheometry has also been used for measuring the dynamic viscoelastic properties of cells.
Magnetic beads attached to cells can be manipulated by the external-generated magnetic field,
therefore apply various loadings on cells. The deformation of the cell is imaged by tracking the
non-magnetic beads dispersed on the surface of the cell. However, both AFM and magnetic bead
micro-rheometry are invasive methods [75]. The experimental setup is complex, and the
operation is time-consuming.
All the five methods I discussed above are invasive methods. Mechanical forces are
directly applied on cells, such as AFM tip, micropipette, optical tweezers, magnetic beads or
microplate and then cell deformation is monitored by using optical microscopy or photodiodes.
The dimensions, and the contact between probe and cell surface are significantly affect the
values of shear modulus. Direct contact between probe and cell will alter the local organization
of cytoskeleton, which affects cell mechanical properties. Micro/nano-particle tracking method is
a non-invasive and passive technique for characterization of cell viscoelasticity. In the
measurements, micro or nano-particles are dispersed on the surface of cell membrane or in the
cytoplasm of a cell. The displacement of these particles is mapped by tracking the path of the
particles over time through successive digital images. The Brownian motion of these particles
can be used to extract the viscoelastic properties of materials around them. Therefore, through
tracking the path of micro or nano-particles, the local viscoelastic properties of cells are obtained.
Overall, micro/nano-particle tracking method is a non-invasive, fast and sensitive technique to
quantify viscoelastic modulus of cells. However, in many cases, it is hard to find natural particles
in cells in terms of the location, particle dimension and so on. Based on the current techniques
47
for cell viscoelasticity measurements, it is necessary to develop a simple, fast, non-invasive and
quantitative method to characterize cell viscoelasticity. In this dissertation, we applied acoustic
wave sensing technology, which is a non-invasive method, to characterize cell viscoelasticity.
1.4.5 Cell traction force
When cells attach to and spread on the substrate, they generate internal tensile forces through
actomyosin interaction and actin polymerization and exert traction forces on the substrate or
extracellular matrix (ECM) through focal adhesions [76]. This traction force is very important in
cell shape maintenance and mechanical signal transmission throughout the intracellular space,
therefore it is important for cells to probe and respond to various physical signals at the
extracellular environment. Understanding the regulation of cell traction forces to cell biological
function and the ability to measure cell traction forces are critical in many physiological and
pathological events, such as inflammation, wound healing, embryogenesis and metastasis [76].
Many methods have been developed to measure cell traction forces for both populations
of cells and the single cell, including cell-populated collagen gel, thin silicone membrane, micro-
machined force sensor array and cell traction force microscopy. Cell-populated collagen gel
method is used to measure traction forces generated by population of cells. In the measurements,
cells are embedded into the collagen gel and the traction forces by cells induce the contraction in
gels. Therefore decrease in the diameter of the gel disk can be used to determine cell traction
forces. This is the simplest method to measure the traction forces of the population of cells.
However, this approach does not measure the traction forces of individual cells. It is a qualitative
tool for cell traction force assay. A thin silicone film has been applied to measure cell traction
forces [77]. When cells adhere or move across the film, cell traction force generates wrinkles in
48
the silicone film. These wrinkles are easily detectable with a light microscope and used to
determine the traction forces of individual cells. However, this method has a poor spatial
resolution. Because the non-linear relationship between the forces and the wrinkle geometry is
too complex for facile quantitative analysis, the wrinkling substrate assays remain a qualitative
tool [77]. More recently, two quantitative methods have been developed to measure the traction
forces of the individual cell with a high spatial resolution: force sensor array and cell traction
force microscopy. Force sensor array uses an array of micro-machined cantilever beams or
polymer micropost array [78]. The deflection of the cantilever beam determines cell traction
force. The limitation of this method is that it only can quantify cell traction force in one
direction. The micropost force sensor array fabricated by soft lithography overcomes this
limitation [79, 80]. Each micro-post is an individual force sensing unit and measures cell traction
force applied on it independently. So micropost force sensor array can detect cell traction forces
in all directions. Nowadays, studies on development of micropost force sensor array is still going
on, focused on the fabrication of high density micropost array and image analysis to improve the
sensitivity and reliability of this method. Another quantitative method for measuring cell traction
forces is cell traction force microscopy [81]. It uses elastic polymacrylamide gel (PG) substrate
embedded with fluorescent beads. Cells are cultured on this substrate and deform it. The
deformation in the substrate is determined by tracking the displacement of fluorescent beads
embedding in elastic polymacrylamide gel (PG) substrate with and without cell adhesion to it.
The deformation in the substrate is used to compute cell traction forces. The most important
issues in cell traction force microscopy are how to determine the displacement field of the
substrate and how to compute cell traction force through the substrate deformation. With the
progresses in solving these two problems, cell traction force microscopy now provides the most
49
reliable and comprehensive information on cell traction forces distributed in an entire single cells
[81]. Nowadays, cell traction force microscopy has been used to determine the regulation of cell
traction force in cell behaviors, such as cell migration and specific gene expression [57, 82, 83].
In this dissertation, we used cell traction force microscopy to study the effects of cell shape on
cell traction force.
50
2.0 RESEARCH OBJECTIVE
Because of potential opportunities of cell-based biosensors in fundamental cell biological
research, drug development, environmental pollution, etc, cell-based biosensors have received
more and more attentions. The development of innovative cell-based biosensor systems requires
detailed research on transduction techniques, interface engineering and the integration of
micro/nano-fabrication and biology. As discussed in chapter 1, acoustic wave sensors, especially
thickness shear mode resonator and Love mode acoustic wave device, are suitable as cell-based
biosensors. However, current research on acoustic wave devices as cell-based biosensors is
limited, focused on the thickness shear mode resonator sensors for monitoring cell attachment
and spreading. More detailed research activities, including theoretical modeling and
experimental studies, are required to extend the applications of acoustic wave devices as cell-
based biosensors. Furthermore, for cell-based biosensors, controlling cell function by controlling
living/non-living interface in sensor systems is critical to increase the sensor sensitivity and
reliability since it can reduce the variability of individual cell response and preserve the
physiologic responses of the cell. Therefore, studies on engineering the interface to control cell
function and determining cell response to the engineered interface are important. Currently, soft
lithography has been widely used to control the adhesive protein distributions on the substrate.
Using this technique, people have shown that cell spreading area regulates a lot of cell function.
However, no studies have been performed to determine the effects of cell shape on cellular
51
structure, traction forces and protein synthesis. So, the objective of this dissertation is to develop
acoustic wave sensor systems for cell biological studies and to determine the effect of cell shape
on cell structure, mechanics and function by soft lithography techniques. Based on this overall
objective, this dissertation has the following specific aims:
To apply the TSM device to monitor the processes of cell attachment, spreading and
formation of focal adhesion after cell addition to the medium and detachment procedure
under cytoskeleton disruption (chapter 3).
To characterize the viscoelastic properties of cell monolayer by the TSM sensor (chapter
4).
To develop a theoretical model of wave-guided shear-horizontal surface acoustic wave
devices as cell-based biosensor (chapter 5).
To determine the role of cell morphology in cell-ECM adhesion, cytoskeleton structure
and traction forces, and cell biological function by precisely controlling cell shape using
soft lithography techniques (chapter 6).
In the last chapter of this dissertation, the principle conclusions and impact of this work, as well
as the most promising avenues for future research will be summarized.
52
3.0 MONITORING CELL ATTACHMENT, SPREADING AND FORMATION OF
FOCAL ADHESION BY USING THICKNESS SHEAR MODE ACOUSTIC WAVE
SENSORS
3.1 INTRODUCTION
As discussed in chapter 1, quartz thickness shear mode (TSM) acoustic wave sensor is a
promising method for quantitatively monitoring cell attachment and spreading in real time. In the
last two decades, studies have been performed to monitor this procedure with different types of
cells [20-22, 25]. Their results have successfully demonstrated that resonant frequency shifts and
motional resistance changes are related to the fractional surface coverage by cells. By comparing
different cell lines under otherwise identical experimental conditions, it has been found that each
cell line has a characteristic frequency shift. These studies, however, only provide a qualitative
observation of cell behaviors on the sensor surface, and little is known about the detailed cell
behaviors and properties that are responsible to frequency shift and resistance change during cell
attachment and spreading. It is highly desirable that a theoretical model can be developed to
relate the behaviors and properties of attached cells to TSM sensor electric signals.
In this chapter, the theoretical modeling and experimental studies for monitoring the
processes of cell attachment, spreading and formation of focal adhesion after cell addition to the
medium and detachment procedure under cytoskeleton disruption will be discussed. A one-
53
dimensional multilayer sensor analytical model for cell monolayer adhesion on sensor surface is
proposed to relate resonant frequency and motional resistance to viscoelastic properties of the
cell layer and the interfacial layer between cells and the sensor surface. Besides the case of cell
monolayer adhesion, we also developed a theoretical model to describe the resonant frequency
shift and resistance change due to the non-uniform cell adhesion. Experimentally, the dynamic
processes of cell attachment, spreading and formation of focal adhesion after cell contacting a
surface have been monitored by measuring the resonant frequency shift and motional resistance
change for different cell seeding densities. It has been found that during cell spreading, the
resonant frequency shift and resistance change are proportional to cell seeding densities, which
agrees with our non-uniform cell adhesion model. In addition, by comparing the theoretical
predictions on frequency shift and resistance change with experimental results, it has been shown
that during the formation of focal adhesion, the resonant frequency shift and resistance changes
are responsible to the changes in the viscoelastic properties of cell layer and cell-substrate
distance. Furthermore, knowing that the actin cytoskeleton is important to cell adhesion and the
mechanical properties of cells, we have also investigated the motional resistance change caused
by the disruption of actin cytoskeleton induced by fungal toxin cytochalasin D in the human skin
fibroblasts. Our results have shown the changes in cell-substrate contact and morphology
induced by fungal toxin cytochalasin D decrease the motional resistance.
54
3.2 THEORETICAL CONSIDERATIONS OF CELL-BASED TSM ACOUSTIC
WAVE SENSOR
3.2.1 Generic modeling of quartz TSM sensor with loading
A general approach to theoretically describe the transduction mechanism of acoustic wave
resonator sensors is based on solving a set of wave equations by applying boundary conditions
between different layers to get the input electrical impedance of the resonator [12, 84-86].
Resonant frequency shift or mechanical loss can be correlated to the viscoelastic properties and
thickness of the surface layer or the medium that the resonator contacts. However, it is quite
complex to solve those wave equations, especially in the case of multilayer loadings. Here we
apply transfer matrix method to obtain a generalized electrical impedance of quartz TSM
resonator sensor with multilayer loading on the top electrode. If the electrode exposed to air is
treated as a backing layer with acoustic impedance of LLLL vFZ =1 , the transfer matrix of the one-
dimensional TSM resonator that relates the input electric voltage and current to the multilayer
loading (Figure 22) on the other side of the device is given by
[ ] ⎥⎦
⎤⎢⎣
⎡×=⎥
⎦
⎤⎢⎣
⎡
LR
LR
vF
AIV
(3.1)
where matrix A have the following format:
[ ]⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
+−
++×⎥
⎦
⎤⎢⎣
⎡=
qqq
qqq
qqq
q
o
o
ZZ
jZj
jZZ
ZZZ
j
CjCj
HSA
γγγ
γγγγ
ωωϕ
ϕ sin)1(cos2sin
)sincos(sincos
01
1
112
(3.2)
In this equation, H=cosγq–1+jZ1sinγq/Zq, φ=hCo, h=eq/εq, Co=εqS/hq, γq=ωhq/(cq/ρq)1/2, Zq=(ρqcq)1/2,
eq=e26, εq=ε22, and cq=c66+e262/ε22+jωηq. Co is the static capacitance of piezoelectric quartz, and S
is the area of quartz. c66, e26, ρq, ε22 and ηq are the components of the material property tensors
55
for mechanical stiffness, piezoelectric constant, density, permittivity and viscosity of quartz [87].
If the electrode effect is neglected, since one side of the resonator sensor is exposed to air, we
have . According to Eq. (3.2), the input electrical impedance of the quartz resonator
sensor with acoustic wave loading impedance
01 =Z
LRLRL vFZ = on the top electrode is given by:
⎟⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜⎜
⎝
⎛
−
−
−==q
q
L
q
Lq
qo
ZZj
ZZj
kCjI
VZγ
γ
γω cot1
2tan2
11 2
(3.3a)
where k2=eq2/(εqcq) is the electromechanical coupling coefficient of the quartz.
The abbreviations in equation (3.1) are calculated with the following equations:
q
ce
Kε
22 = (3.2b)
qqq ch /ρωα = (3.2c)
qqq cZ ρ= (3.2d)
qq h
AC ε=0 (3.2e)
26eeq ≡ (3.2f)
22εε ≡q (3.2g)
qq jecc ωηε
++≡22
226
66 (3.2h)
where K2 is the electromechanical coupling coefficient of quartz, α is the acoustic phase shift
inside the quartz crystal and Zq is the characteristic acoustic impedance of the quartz crystal. C0
is the static capacitance of piezoelectric quartz, and A is the area of quartz. c66, e26, ρq, ε22, hq and
ηq are the components of the material property tensors for mechanical stiffness, piezoelectric
56
constant, density, permittivity, thickness and viscosity of quartz. ZL is acoustic loading
impedance, defined as the ratio between the stress and displacement at the sensor surface.
I V
v3
F3
v2
F2
vLR
FLR
VLL
FLL=0
B1 B2A1 Vn+1
Fn+1=0
vn
Fn
Bn
Figure 22. Transfer matrix model of a thickness shear mode (TSM) quartz resonator with multiple uniform layers loaded on one side
The electrical impedance can be separated into a parallel circuit consisting of the static
capacitance Co and motional impedance Zm. From Eq.(3.3), the motional impedance Zm can be
written as follows:
)2
tan(21
14
11)
2tan(2
12
2
Lq
Lq
oq
q
om
Z
ZjZZ
kCk
CjZ
γ
γωγ
γ
ω −+
⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜
⎝
⎛
−=
(3.4)
The motional impedance is split into two parts, and , i.e., the motional electrical
impedance of unloaded quartz resonator and impedance change caused by uniform acoustic
loading:
omZ L
mZ
57
⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜
⎝
⎛
−= 1)
2tan(2
1 2
q
q
o
om
kCj
Z γ
γ
ω
(3.5a)
)2
tan(21
14
12
Lq
Lq
o
Lm
Z
ZjZZ
kCZ
γ
γω −
=
(3.5b)
By taking the approximation of tan(γq/2)=4γq/[(Nπ)2–γq2] near the resonant frequency, (N
is the harmonic number), the motional impedance of unloaded quartz near the series resonant
frequency is expressed as
qqq
qqoq
om RLj
CjRLj
CjCjZ ++
′=++
−+= ω
ωω
ωω111
(3.6)
The circuit parameters in the motional impedance have been given by Lucklum et al
[85, 88-90] and are listed here for convenience, qoq
qq ch
SeC 2
28π
=,
22'
81 πkC
C qq −
=,
2
3
22
2
88 q
ooq Se
hkC
Lρ
ωπ
==,
qoo
q
q
qoqq ckCC
cR 2
2
8πηη
==, and 22
226
66 εe
ccqo +=.
For small uniform loading, ZL/Zq<<2tan (γq/2), Eq. (3.5b) can be further simplified and
can be expressed as, LmZ
q
Lq
o
Lm Z
ZkC
Z24
1 γω
= (3.7)
According to Eqs. (3.6) and (3.7), a direct relation between impedance of the acoustic
loading, ZL, and the frequency shift at the maximum of the in-phase electrical admittance as well
as the changes in the motional resistance are obtained, shown in Eqs.(3.8a) and (3.8b) (Lucklum
and Hauptmann, 2003).
58
q
L
o
s
ZZ
ff
π)Im(
−=Δ
(3.8a)
q
L
qo
s
ZZ
LR
πω)Re(
2=
Δ
(3.8b)
3.2.2 Acoustic impedance of uniform multilayer loading and non-uniform loading
For multilayer loading with continuous boundary conditions between each layer, a transfer
matrix model of quartz TSM sensor with multilayer loading can be created (Figure 22). Here B1,
BB2, ... , Bn are transfer matrixes of each non-piezoelectric uniform loading layer. The transfer
matrix for a non-piezoelectric layer B is given by,
[ ] ⎥⎦
⎤⎢⎣
⎡×⎥
⎦
⎤⎢⎣
⎡=⎥
⎦
⎤⎢⎣
⎡×=⎥
⎦
⎤⎢⎣
⎡
i
i
i
i
o
o
vF
ZjjZ
vF
BvF
γγγγ
cossinsincos
(3.9)
where F and v represent stress and particle velocity, respectively; subscript i and o represent
input and output ports of the layer; Z and γ are the characteristic acoustic impedance and the
phase delay of the layer [87]. With these transfer matrixes, the overall acoustic impedance of
uniform multilayer loading ZL can be calculated using Eq.(3.10):
[ ] [ ] [ ] ⎥⎦
⎤⎢⎣
⎡××⋅⋅⋅××=⎥
⎦
⎤⎢⎣
⎡
+
+
1
121
n
nn
LR
LR
vF
BBBvF
(3.10)
where Fn+1=0, since the top layer is exposed to air.
It should be noted that the above discussion is for the case of uniform multilayer loading
on the entire sensor surface. In the application of the sensor system, however, the loading is often
non-uniform in thickness. Lucklum and Hauptmann recently developed a model for sensors with
non-uniform loading. In this model, the sensor is divided into several subunits and a uniform
loading is applied on each subunit [90]. This model can be represented as an equivalent circuit in
59
which each subunit has its own motional impedance in parallel to the static capacitance Co and
all other motional impedances (Figure 23). According to Eqs. (3.5a) and (3.5b), the motional
admittance of each subunit is proportional to Coi therefore proportional to the area of each
subunit Si. The total motional electrical impedance of the resonator with non-uniform loading
can be expressed as
∑
= +
= n
iLm
om
oi
totalm
iZZSS
Z
1
/1
(3.11)
where is the motional impedance of the unloaded resonator, omZ iL
mZ is motional impedance
change if the loading on subunit i is uniformly applied onto the entire surface of the resonator
and S is the area of the resonator electrode. Therefore, the motional electrical impedance change
induced only by the non-uniform loading is given by
o
omn
iLm
om
oi
Lm Z
ZZSS
Z
i
total −
+
=
∑=1
/1
(3.12)
C0
Zm01
ZmL1 Zm
Ln
Zm0n
Figure 23. Equivalent circuit of TSM resonator with non-uniform loading
60
3.2.3 Physical model of cell adhesion onto the sensor surface and its acoustic loading
impedance
Cells attach to the underlying substrate through complex transmembrane receptors, which
connect the cytoskeleton to the extracellular matrix [91]. When cells are steadily adhered to the
substrate, there exists an interfacial gap between cell basal membrane and substrate. The values
of the interfacial gap vary with different contact types. According to surface plasmon resonance
microscopy (SPRM) studies [92], the gap in focal contact area is only a few nanometers, while
cell membrane is separated from the substrate about 30nm in close contact area and about 100-
150 nm in the rest part of cells, respectively. Furthermore, it has been shown that the properties
of cell medium on top of cell layer have minimal effect on the frequency shift and resistance
change since the cell layer is relatively thick and the acoustic wave decays before reaching the
cell medium [23]. Based on these considerations, it is reasonable to include 2 layers on the TSM
resonator sensor surface: a finite interfacial gap and a semi-infinite cell layer. In addition, for
simplicity, all the three layers are assumed uniform in the cell-covered area (Figure 24).
TSM Sensor
MediumCell layer (hcl)
Interfacial layer (hil)
TSM sensor (hq)TSM Sensor
MediumCell layer (hcl)
Interfacial layer (hil)
TSM sensor (hq)
(a) (b)
Figure 24. (a) cell monolayer adhering to TSM sensor surface; (b) multilayer model of cell monolayer on sensor surface
61
For the case that cells attach on the entire surface of the resonator, the total acoustic loading
impedance caused by the uniform interfacial layer and cell layer is given by,
)tan(
)tan(
ilil
ilclclilil
ilil
ilililclcl
ililL
hG
GjG
hG
GjGGZ
ρωρρ
ρωρρρ
+
+=
(3.13)
where ρ, G and h are the density, complex shear modulus and thickness, respectively. Subscript il
and cl represent the interfacial layer and cell layer respectively.
TSM sensor
Medium1 2 3
TSM sensor
Medium1 2 3
Figure 25. Non-uniform cell adhesion to TSM sensor surface
For the case that cells only attach on the partial surface of the resonator, we treated the
loading as non-uniform loading (Figure 25). In this case, we subdivided the sensor system into
several subunits. On each subunit, the loading is uniform. For example, in figure 25, there are
three subunits. The loading on subunits 1 and 3 are cell monolayer adhesion (a finite interfacial
layer plus a semi-infinite cell layer), while that on subunit 2 is the semi-infinite medium. We can
use an equivalent circuit shown in figure 23 to describe this model, in which each subunit has its
own motional impedance in parallel with the motional impedance of all other subunits and
external capacitance of the sensor. If the cell coverage ratio on the electrode area is A, according
62
to Eqs. (3.6), (3.7), (3.12) and (3.13), the total motional electrical impedance change induced by
cell adhesion on partial resonator surface is
omL
mom
Lm
om
Lm Z
ZZA
ZZAZ total −⎟⎟
⎠
⎞⎜⎜⎝
⎛
+−
++
=−1
21
1
(3.14)
where ,omZ 1L
mZ and are the motional electrical impedance of the unloaded resonator,
motional electrical impedance change induced by cell adhesion, and motional electrical
impedance change by semi-infinite medium loading on the entire resonator surface, respectively.
Further simplification of Eq.(3.14) leads to
2LmZ
2
2
21
21
)1(1
Lm
Lm
om
Lm
Lm
Lm
LmL
m Z
ZZZZA
ZZAZ total +
+−
−+
−=
(3.15)
Since omZ >> 1L
mZ and , Eq. (14) can be simplified as 2L
mZ
221 )( Lm
Lm
Lm
Lm ZZZAZ total +−= (3.16)
Eq. (16) indicates that the difference between and is proportional to the cell coverage
A.
totalLmZ 2L
mZ
3.3 MATERIALS AND METHODS
3.3.1 Experimental setup
The experimental setup of TSM resonators for functional biosensors is schematically shown in
Figure 26. Commercially available AT-cut quartz resonators (International Crystal
63
Manufacturing, Oklahoma City, OK) with the fundamental resonant frequency of 10MHz were
used in this study. TSM resonators were inserted into a standard cell culture plate, which was
placed in an incubator during the measurements. The measuring chamber was formed by putting
a poly(dimethylsiloxane) (PDMS) tube, which can be self-sealed on the quartz surface, onto each
resonator. The two electrodes were connected to an impedance analyzer (Agilent 4294A
Precision Impedance Analyzer, Agilent Technologies, Palo Alto, CA) to measure the impedance
spectra of the resonators. Prior to each experiment, the quartz crystal sensor device was sterilized
with ethanol for half an hour and then washed with de-ionized water for three times. After that,
they were dried with a stream of pressurized air and exposed to UV light for about 1 hour to
sterilize the chamber.
Incubator
Computer
Medium
Cell layer
TSM Resonator
Agilent 4294A Precision Impedance Analyzer
PDMS tube
Figure 26. Experimental setup of thickness shear mode resonators for cell-based functional biosensor
64
3.3.2 Cell culture
Human skin fibroblasts were obtained from the American Type Culture Collection (ATCC,
Manassas, VA 20110, USA) and grown in DMEM supplemented with 10% fetal bovine serum
and 1% penicillin/streptomycin (Invitrogen, CA) in 5% CO2 at 37oC and 100% humidity in
culture flask. The medium was changed twice a week. Once fibroblasts reached confluence, cells
were sub-cultured by the standard trypsinization and centrifugation techniques. The passage of
cells was from 7 to 10.
3.3.3 Monitoring cell attachment, spreading and formation of focal adhesion
The attachment and spreading of human skin fibroblasts were monitored after suspended cells
were added onto the resonator surface. Prior to the deposition of cells, the electrical impedance
spectra of uncoated resonators were measured. Cells were then trypsinized, centrifuged and
counted with a hemacytometer. Cell-free medium was loaded into one sensor well as a control, at
the same time medium containing cells with numbers of 7.5×104, 5×104, 3×104, and 1×104 was
each added into the other sensor wells. Cell solutions were also added onto gold-coated glass
slides under identical conditions as a simulation experiment. The electrical impedance spectra for
each resonator sensor were recorded continuously. Images of cells on the gold-coated glass were
recorded on a microscope with a CCD camera.
65
3.3.4 Cells with cytochalasin D (CD) treatment
20 hours after cells cultured on the resonator surface, cells were exposed to CD with
concentration of 0 μM, 2.5 μM, 5 μM and 10 μM. The motional resistance for each resonator
sensor was recorded continuously to monitor correspondingly the response of resonator sensors
to CD treatment.
3.3.5 Immunostaining of actin filaments and nucleus
Cells treated with 5 uM CD for 3, 5, 10, 20 and 40 min were fixed with 4% paraformaldehyde
(FD NeuroTechnologies, Ellicott City, MD) for 20 min, pretreated with 0.1% Triton-X100 for 2
min, and blocked with 5% bovine serum albumin for 1 h. The TRITC conjugated phalloidin was
then applied at a dilution ratio of 1:100 for 45 min. The sample was washed extensively with
PBS to remove excessive phalloidin. The DAPI was applied at a dilution ratio of 1:5000 for 1
min and then sample was washed extensively with PBS again. After that, the samples were
examined with fluorescence microscopy (Eclipse TE2000-U, Nikon, Japan).
3.4 RESULTS
3.4.1 Modeling results for cell adhesion
The relative resistance change (ΔRr) and frequency shift (–Δfr) of TSM sensors due to cell
attachment and adhesion with different cell-substrate contact as compared to cell free medium
66
can be simulated using the theoretical model. Since ΔRr and –Δfr are proportional to cell
coverage, it is sufficient to simulate the resistance and frequency responses for full cell coverage
on the resonator surface. For simplicity, it is assumed that the viscoelastic property of the
interfacial layer between cell and sensor substrate is the same as the cell medium (water). The
resistance change ΔRr as a function of the thickness (hil) of interfacial layer at different viscosity
and shear storage modulus of cell layer is calculated and plotted in Figure 27(A) and (B),
respectively. It can be found that ΔRr increases with decreasing hil from 300nm to 0. If a cell
storage shear modulus value of 1Kpa is used, ΔRr increases with increasing the viscosity of cell
layer (Figure 27 (A)). If a cell viscosity value of 2cp is used, ΔRr decreases with increasing the
storage shear modulus of cells when hil is greater than 90nm but increases when hil is less than
90nm. Changes in storage shear modulus of cells do not induce significant resistance change
(Figure 27 (B)). The frequency shift (–Δfr) as a function of interfacial layer thickness is plotted
in Figure 27 (C) and (D) at different viscosity and shear storage modulus of the cell layer,
respectively. (–Δfr) decreases first and then increases with the decrease of the thickness of
interfacial layer hil from 300nm to 0. For storage shear modulus value of 1KPa, (–Δfr) decreases
with increasing the viscosity of cells when hil is greater than 60nm but increases when hil is less
than 60 nm (Figure 27 (C)). For a cell viscosity value of 2cp, (–Δfr) always decreases with
increasing the shear storage modulus of cells (Figure 27 (D)). It should be noted during the
course of cell attachment and spreading, the thickness and property of interfacial layer and
viscoelastic property of the cell layer are dynamic parameters. Clearly, the resistance change and
frequency shift can be used as quantitative indications of the cell attachment and spreading on
the sensor surface.
67
Figure 27. Modeling results of relative motional resistance increase and resonant frequency decrease to cell-free medium due to cell monolayer adhesion. (A) Relative resistance increase with changing thickness of interfacial layer
and viscosity of cell layer when the storage shear modulus of cell layer is fixed to 1KPa; (B) relative resistance increase with changing thickness of interfacial layer and storage shear modulus of cell layer when the viscosity of cell layer is fixed to 2cp; (C) relative frequency decrease with changing thickness of interfacial layer and viscosity of cell layer when the storage shear modulus of cell layer is fixed to 1KPa; and (D) relative frequency decrease with
changing thickness of interfacial layer and storage shear modulus of cell layer when the viscosity of cell layer is fixed to 2cp.
3.4.2 Process of cell adhesion after cells contact with a surface
The morphology of cells cultured on the gold-coated glass surface with different cell seeding
densities was recorded at different culture time points using microscopy with CCD camera
68
(Figure 28(A)). For all seeding densities, cells are with spherical shapes and suspend in the
medium a few minutes after the cell additions. Then, most of cells are found to attach onto the
resonator surface during the first 10 min and then start spreading. For high seeding densities (cell
number: 7.5×104 and 5×104), cells cover the entire resonator surface about 50 min after the cell
addition. However, some cells do not have enough places to attach and spread at that time point.
After that, the rearrangements occur and cell monolayer is formed finally at 500min. For the
intermediate (cell number: 3×104) and low seeding density (cell number: 1×104), after cells
attach to the surface, they spread continually until they establish cytoskeleton network and reach
a steady adhesion state. During their spreading, there is no obvious rearrangement observed.
During the course of cell attachment, spreading and formation of a steady state of cell
adhesions, both (ΔRr) and (–Δfr) experience four phases: fast increase stage, slow increase stage,
decrease stage and steady state stage (Figure 28(B) and (C)). At the decrease stage, ΔRr
decreases much faster for high densities than the intermediate seeding density. There is only a
very small decrease in (ΔRr) at the low seeding density. However, (–Δfr) decreases rapidly for all
seeding densities at the decrease stage. In addition, (ΔRr) and (–Δfr) are dependent on cell
seeding densities at fast increase stage except the highest one, while they are not significantly
affected by seeding densities at the steady state stage.
69
5min 15min 50min 200min 500min
1X104
3X104
5X104
7.5X104
A
Figure 28. (A) Images of cells in culture at different time points; (B) time course of the relative motional resistance increase; and (C) relative resonant frequency decrease to the cell-free medium after different numbers of
human skin fibroblasts seeded onto the resonator surface at time point zero
3.4.3 Cell layers with CD treatment
Cytochalasin D is a fungal toxin that severely disrupts network organization, increases the
number of actin filament ends, and leads to the formation of filamentous aggregates (Schliwa,
70
1982). The actin stress fibers are distributed evenly and the actin filaments form end-to-side
associations with other actin filaments for untreated cells (Figure 29(A)). However, after cells
are treated with 5μM cytochalasin D, the filamentous network is rather non-uniformly distributed
throughout the cell. The dense network aggregates form first in the cell periphery (Figure 29(B),
(C) and (D)), and later throughout the cell body (Figure 29(E) and (F)). With a treatment longer
than about 20 min, cells show the dissolution of stress fibers.
For 5μM Cytochalasin D treatment, motional resistance has a significant change after
about 7-8 minute treatment as compared with the control experiment, and continuously shifts
during the first hour. In addition, the values of resistance decrease after about 7-8 minutes are
dosage dependent (Figure 29(G)).
71
A
B
C F
E
D
Figure 29. Florescence images of human skin fibroblasts exposed to 5uM cytochalasin D for different time: (A) 0min; (B) 3min; (C) 5 min; (D) 10 min; (E) 20 min; (F) 40min; and (G) changes in motional resistance after
confluent human skin fibroblast layers treated with Cytochalasin D with different concentrations
72
3.5 DISCUSSIONS
Based on the modeling results, three factors are regarded to significantly affect the frequency and
resistance responses during the cell attachment, spreading and formation of focal adhesions: cell-
covered area, interfacial layer between the basal membrane and substrate, and cell mechanical
properties. When a cell contacts with a surface, cell attachment occurs first, involving the
interaction between integrins and substrate. Then cells increase their surface contact area with
the substrate through formation of actin microfilaments and cell spreading [60]. It is observed
under microscopy that cells attach onto the resonator surface and spread during the first hour of
cell culture. Therefore, there are an increase in cell coverage on the sensor surface and a decrease
in the thickness of interfacial layer. The measurement results indicate that both –Δfr and ΔRr
increase quickly. These experimental observations are consistent with our simulation results that
–Δfr and ΔRr increase with increasing cell coverage and decreasing hil. In addition, at the fast
increase phase, both ΔRr and –Δfr are seeding-density dependent except the highest one (cell
number: 7.5×104). The reason is that the cell-substrate contact and cell mechanical properties are
similar for different seeding densities during this period, therefore ΔRr and –Δfr are proportional
to cell coverage, which is linearly dependent on the seeding densities before the coverage
reaches 100%.
Cell images by microscopy show that 1 hour after cell addition, significant
rearrangements occur because some cells cannot find a space to attach during the first hour for
high cell seeding densities but no rearrangements at the low cell seeding density. It is believed
that cell migration, and decrease in cell spreading area in order to allow more cells attaching on
the surface during cell rearrangements increase the cell-substrate separation, which decreases –
Δfr and ΔRr according to the simulation results. Consistently, it is observed that there is a
73
significant decrease in –Δfr and ΔRr measurements at high cell densities. However, for the low
seeding densities (cell number: 1×104) it has been found that –Δfr decreases after cells cultured
onto the resonator surface for 2 hours but no such a decrease appears in ΔRr. This observation
may be explained by examining the cell adhesion mechanism. Following spreading, cells
proceed to organize their cytoskeleton as characterized by the formation of focal adhesions and
actin-containing stress fibers. It is believed that the formation of actin stress fibers affects cell
viscoelasticity, especially significantly increasing the storage shear modulus [67, 93]. The
simulation results show that increasing storage shear modulus clG′ slightly increases ΔR when hil
is less than 90nm, but slightly decreases ΔR when hil is greater than 80nm. The effect of clG′ on
ΔRr is not significant. However, increasing clG′ significantly decreases –Δf. So the different
responses of ΔR and –Δf after cells cultured onto the resonator surface for 2 hours at the low
seeding densities might be due to the increase in the storage shear modulus of cells. Therefore,
the TSM resonator sensors can be used to detect the changes in cell viscoelastic properties,
which plays an important role in cell migration, cytokinesis and cell mechanics [55, 67, 94].
The steady values of ΔRr and –Δfr are not dependent on cell number. One possible reason
is that the cell coverage is not proportional to cell number at the steady adhesion stage.
Microscopic images at culture time point of 500 min have indicated that the spreading area of
cells is different for different seeding densities. The lower the cell seeding density, the larger the
individual cell spreading area is. In addition, the cell-substrate separation and cell mechanical
properties could be different for different seeding densities, therefore the differences in the
steady values of ΔRr and –Δfr for different cell seeding densities are not only affected by cell
coverage, but also by cell mechanical properties and the thickness of the interfacial layer. An
interesting experimental observation at the steady adhesion stage is that for high cell seeding
74
densities (cell number: 7.5×104 and 5×104), the steady value of ΔR for cell number 5×104 is
larger than cell number 7.5×104. Since the cell coverage is similar (100% coverage) for both
cases, the larger steady value of ΔR for cell seeding density of 5×104 could be caused by a
smaller cell-substrate separation and a larger value of storage shear modulus of cells.
Cells adhere to the substrate via adhesion proteins whose intracellular domain binds to
cytoskeleton. Without attachment to the cytoskeleton, a cell adhesion protein could be
hydrolyzed by extracellular hydrolytic enzymes and ripped out from the cell membrane.
Therefore, cytoskeleton plays an important role in maintaining strong adhesion between cells and
substrate. Since cytoskeleton is a three-dimensional network extending throughout the cytoplasm,
it is considered as the determining factor for cell mechanical properties, especially the elastic
part [66]. Fungal toxin cytochalasin D disrupts the actin cytoskeleton, as was shown by the
fluorescent images. According to the simulation results, the resistance change is significantly
affected by the thickness of interfacial layer, cell-covered area and viscosity of cells, but is less
sensitive to the storage shear modulus. Therefore, the decrease in resistance after cytochalasin D
treatment can be attributed to the loss of strong adhesion of cells with the substrate and the
subsequent cell shape changes due to the cytoskeleton disruption.
3.6 CONCLUSIONS
In summary, a multilayer resonator sensor model was proposed to establish a relationship
between the resonant frequency and resistance change of the TSM resonator induced by cell
adhesion and spreading and the physical properties of the interfacial layer and the cell layer. By
dividing the resonator into several subunits, we have first time theoretically shown that the
75
relative frequency and resistance changes to cell-free medium due to cell adhesion is linearly
dependent on the cell coverage, which has also been observed by previous experimental studies.
Experimentally, the dynamic processes of cell adhesions as functions of cells seeding densities
have been investigated. By comparing the theoretical prediction on ΔRr and –Δfr using the model
with experimental results, the mechanism that is responsible for the changes in frequency and
resistance during cell attachment, spreading and formation of strong adhesions has been
discussed. Cell rearrangements involving cell migration at high seeding densities cause a large
decrease in resistance and increase in frequency. The different responses of frequency shift and
resistance change during the formation of actin stress fiber in cells indicate the changes in cell
mechanical properties. Changes in cell-substrate contact and morphology induced by fungal
toxin cytochalasin D decrease the motional resistance. This study has demonstrated that the
thickness shear mode (TSM) acoustic wave resonator attached with living cells can be used, in
real time, as an effective functional biosensing device to monitor the process of cell adhesion and
spreading onto a surface, which are important in physiological and pathological processes [95]. It
may also be used to measure cell mechanical properties and interfacial layer characteristics
between cells and the substrate.
76
4.0 THICKNESS SHEAR MODE RESONATORS FOR CHARACTERIZING
VISCOELASTIC PROPERTIES OF MC3T3 FIBROBLASTS MONOLAYER
4.1 INTRODUCTION
In chapter 3, we developed theoretical models of the thickness shear mode resonator sensors with
cell monolayer adhesion and non-uniform cell adhesion on their surface. Experimentally, using
thickness shear mode sensors, we investigated the dynamic procedure of cell attachment,
spreading and formation of focal adhesion with different cell seeding densities. In addition, we
monitored the changes in cell adhesion under cytoskeleton disruption. Our theoretical model and
experimental studies have demonstrated that electrical signals of TSM resonator (e.g. resonant
frequency, motional resistance, and electric impedance or admittance) are related to the
viscoelastic properties of cell layer when they attach to the sensor surface. So in this chapter, we
investigated the feasibility of characterizing cell viscoelasticity using TSM acoustic wave
sensors.
Cell viscoelasticity, determined by lipid membranes and cytoskeleton, plays an important
role in regulating many cell behaviors. A high-precision measurement of the cell viscoelasticity
provides critical information to quantify the effect of drugs, mutations or diseases on cell
structures [65]. As discussed in chapter 1, up to now, many techniques have been developed to
measure the viscoelastic properties of cells, including atomic force microscopy (AFM) [96],
77
magnetic bead Microrheometry (MBM) [65, 97], micropipette [98], microneedles [99] and cell
poker [100]. However, all of the above techniques directly apply mechanical forces on cells,
therefore they are invasive methods. In addition, the system is complex and measurement is also
time-consuming. Laser tracking microrheology, a non-invasive, fast and sensitive technique, was
developed to quantify viscoelastic modulus of cells recently. Endogenous cellular particles were
used as probes. However, it is hard to obtain the natural particles in cells that meet high
requirements in terms of the location or dimensions [64]. Therefore, it is highly desirable to
develop a non-invasive, simple and fast tool to characterize cell viscoelasticity. So in this chapter,
we characterized the viscoelastic properties of MC3T3 fibroblast momolayer with TSM quartz
resonators. In order to obtain comprehensive and detailed information on the resonator
oscillation, the impedance over a range of frequencies near resonance were measured by the
impedance analyzer instead of only measuring the resonance frequency and motional resistance
with an oscillation circuit. After cells were fully spread to form a monolayer on the surface, the
viscoelastic properties of the cell monolayer were evaluated by fitting experimental admittance
curves with those obtained from the theoretical model. Compared with other techniques, our
results demonstrated that TSM acoustic wave sensor provides a reliable tool to characterize cell
viscoelasticity in a non-invasive, fast and simple way.
78
4.2 MATERIALS AND METHODS
4.2.1 Experimental setup
In the experimental setup of TSM resonators for functional biosensors (Figure 26), the AT-cut
quartz resonators with the fundamental frequency of 5MHz are commercially available (Stanford
Research Systems Inc., Sunnyvale, CA). TSM resonators were inserted into a standard cell
culture plate, which was placed in an incubator during the measurements. The measuring
chamber was formed by putting a poly(dimethylsiloxane) (PDMS) tube, which can be self-sealed
on the quartz surface, onto each resonator. The two electrodes were connected to an impedance
analyzer (Agilent 4294A Precision Impedance Analyzer, Agilent Technologies, Palo Alto, CA)
to measure the admittance spectra of the resonators. Prior to each experiment, the quartz crystal
sensor device was sterilized with ethanol for half an hour and then washed with de-ionized water
for three times. After that, they were dried with a stream of pressurized air and exposed to UV
light for about 1 hour to sterilize the chamber.
4.2.2 Cell Culture
MC3T3-E1 cells (ATCC, Manassas, VA) were grown in the α-MEM growth medium (Invitrogen,
Carlsbad, CA) supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA),
100units/ml penicillin, and 100μg/ml streptomycin (Sigma, St. Louis, MO). The cells were
maintained at 37oC in a humidified atmosphere of 95% air and 5% carbon dioxide. For all
experiments, cells were plated on the sensor surface at an initial density of 2Х104 cells/cm2.
79
4.2.3 Measurements
Prior to the deposition of cells, the electrical admittance spectra and total parallel capacitance of
the uncoated resonators and water-loaded resonators were measured first. Then the electrical
responses of resonators with α-MEM growth medium solution were measured to characterize the
properties of the solution. 24 hours after cell-free medium and cell medium were added onto
each resonator surface, the admittance spectra and the total parallel capacitance of each resonator
sensor were recorded.
4.3 DETERMINATION OF THE VISCOELASTIC PROPERTIES OF CELL
LAYERS
According to Eq. (3.3), without taking any approximation, the input electrical impedance of
TSM resonators with loading on one side can be expressed as an equivalent circuit consisting of
a static capacitance C0 parallel to motional impedance Zm. Because the fixture and cable induce
an external capacitance that has to be considered here, the generalized input electric admittance
of resonator sensors is rewritten as
mZ
CjjBGY 1*0 +=+= ω (4.1)
where C0* is the total parallel capacitance of resonators, which is the sum of the static
capacitance C0 and the external parallel capacitance Cex, and Zm is the impedance of the motional
branch [25]. The motional impedance can be rewritten based on Eq. (3.3):
80
⎟⎟⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜⎜⎜
⎝
⎛
−
⎥⎥⎦
⎤
⎢⎢⎣
⎡−
−
= 1
2tan2
cot11
20
q
L
q
L
m
ZZ
jK
ZZ
j
CjZ
αα
α
ω
(4.2)
As discussed in chapter 3, the total acoustic loading impedance caused by the uniform interfacial
layer and cell layer is given by
)tan(
)tan(
ilil
ilclclilil
ilil
ilililclcl
ililL
hG
GjG
hG
GjGGZ
ρωρρ
ρωρρρ
+
+= (4.3)
where ρ, G and h are the density, complex shear modulus and thickness respectively. Subscript il
and cl represent interfacial layer and cell layer respectively.
4.3.1 Generic consideration for determining the viscoelastic properties of cell monolayer
According to Eq. (4.2), the input admittance of the resonator with loadings is determined by the
quartz parameters (e26, ε22, c66, ρq, hq, A and ηq) and the surface acoustic loading impedance ZL.
Of all quartz parameters, e26, ε22, c66 and ρq can be obtained from the literatures, while hq, A and
ηq must be calculated by sensor calibration because these parameters are affected by the
electrodes, fixtures et al. ZL is the only ‘foreign parameter’ in Eq. (4.1), which are dependent on
the material properties of loading layers shown in Eq. (4.3). Therefore, the input admittance of
the resonator can be calculated when the properties of the loading layers and quartz parameters
are known. However, for the resonators as sensors, we have to extract the properties of the
loading layers from the measurement of the electric admittance spectra. For the cell monolayer
on the sensor surface, the sensor structure includes the quartz resonator, finite interfacial layer
and semi-infinite cell layer. In this three-layer model, there are seven unknown parameters in
81
acoustic loading impedance. Four of them correspond to the interfacial layer (ρil, hil, Gil' and Gil'')
and three to the cell layer (ρcl, Gcl' and Gcl''). In these seven parameters related to loading layers,
the density and shear modulus of interfacial medium layer can be characterized by loading the
cell culture medium alone onto the resonator surface and furthermore we assume the density of
cell monolayer is the same as water. Therefore, the unknown parameters can be reduced to three:
hil, Gcl', and Gcl''.
4.3.2 Algorithmic strategy of exacting the viscoelastic parameters of cell layer
The algorithmic strategy of extracting parameters of loading layers on TSM resonators from the
measurement of the electric admittance spectra has been developed recently by Yolanda Jimenez
et al.[101], which is used for extraction of the viscoelasticity of cell monolayer in this
dissertation. Basically, there are three steps: sensor calibration, interfacial layer characterization,
and calculating the shear modulus of cell layers (Figure 30). In the following, I will briefly
describe these three steps.
82
G2’j and G2
”j and h1j
Y_water(ω) C0' (10MHz)
ηq, hq, C0 and Cex
Sensor
Y_medium(ω1) C0' medium (10MHz)
Senor calibration
Z1c
Calculating the viscoelastic properties of cell layers
Y_cell adhesion (ω) C0'_cell adhesion (10MHz)
ρ1
Interfacial layer characterization
Figure 30. Flow chat for exacting the properties of loading layers from the measurements of the electric admittance spectra of TSM resonators.
Sensor calibration
As mentioned, before extracting the viscoelastic parameters of cell layers, a calibration
process is carried out to calculate the effective TSM resonator’s parameters. The inputs include
electric admittance spectrum of the resonator sensor only in contact with water, the total parallel
capacitance and the characteristic impedance of water. The outputs are the effective quartz
parameters, including the quartz’s thickness (h
*0C
q), viscosity (ηq), static capacitance (C0) and
external capacitance (Cex). The calibration process is as follows: (a) The admittance spectrum
Y(ωi) of the water-loaded resonator is measured by using the impedance analyzer (where i=1 to
83
M with M the number of frequency points). C0* is measured at two times of fundamental
resonant frequency. Zm(ω0) is then obtained from Eq. (4.1). (b) In a reasonable range, ηq is
discretized into ηqj (where j=1 to n with n is the number of ηq). At first, a high range is tried.
Then it can be reduced to a lower range for precision requirement after several tries. (c)
According to Eq. (4.2), C0 can be expressed as
( ) ⎟⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜⎜
⎝
⎛
−
⎥⎥⎦
⎤
⎢⎢⎣
⎡−
−
= 12tan2
cot11
20
q
L
q
L
m
ZZjK
ZZj
ZjC
αα
α
ω
(4.4)
For each Zm(ω0) and ηqj, the thickness hq
j is obtained through canceling the imaginary part of
Eq.(4.4). A Regula-Falsi based algorithm is applied for getting the values of . For each (ηqh qj, hq
j),
the corresponding values of C0j are obtained from Eq. (4.4). (d) The solution of parameters (ηq,
hq, C0) is obtained as the one that make the smallest admittance spectrum error between
theoretical and experimental values. The error function is defined as follow:
∑∑
∑∑
==
==
+
−+−= M
iEXPi
M
iEXPi
M
i
MODEL
jiEXPi
M
i
MODEL
jiEXPi
jBG
BBGGError
1
2
1
2
1
2
1
2
))(())((
))()(())()((
ωω
ωωωω (4.5)
Interfacial layer characterization
According to our model, there is an interfacial layer between cells and the substrate. We
assume that the properties of this interfacial layer are the same as those of the cell culture
medium. Therefore, for interfacial layer characterization, the inputs include the calibration
parameters obtained from the sensor calibration, and the electrical admittance spectrum and total
parallel capacitance of the resonator loaded with the cell culture medium only. According to Eqs.
(4.1) and (4.2), the acoustic impedance of medium loading ZL can be easily calculated. Since
84
ZL=(jρilGil'')1/2 and the density of cell culture medium ρil is known from the literature, the values
of Gil'' are obtained. Gil' is equal to zero if the interfacial layer is assumed as a Newtonian liquid
layer. Therefore, finally, we obtain ρil, Gil' and Gil''.
Calculating the viscoelastic properties of cell monolayers
The calculation of the shear storage and loss modulus of cell layer is similar with the
sensor calibration procedure [101]. From Eq.(4.3), we can get
[ )2()ln(2
)( πθω
ρρω kjr
jG
hm ilililils ++−== ] (4.6)
where
EXPLilil
EXPLilil
clclilil
clclilil
ZGZG
GGGG
r)()()(
)(ωρωρ
ρρρρ
ω+
−×
+
+= (4.7a)
))()()(
()(EXPLilil
EXPLilil
clclilil
clclilil
ZGZG
GGGG
phaseωρωρ
ρρρρ
ωθ+
−×
+
+= (4.7b)
with -π≤θ≤π, and k=0, ±1, ±2, ….
The extraction process includes the following steps:
(a) The admittance spectrum Y(ωi) of the resonator with cell monolayer adhesion, is measured by
using the impedance analyzer (where i=1 to M with M the number of frequency points). C0* is
measured at two times of fundamental resonant frequency. ZL(ωi) .EXP is obtained from Eqs׀
(4.1) and (4.2).
(b) In a reasonable range Gcl' is discretized into Gclj' (where j=1 to n with n is the number of Gcl').
85
(c) At ω=ω0 (angular resonant frequency), for each Gclj', Gcl
j'' is obtained through canceling the
imaginary part of ms(ω0). For each (Gclj', Gcl
j''), the corresponding values of hilj are obtained from
Eq. (4.6).
(d) The solutions of Gclj', Gcl
j'' and hilj with the value of Eq.(4.4) over 0.5% are ignored.
4.4 RESULTS AND DISCUSSIONS
The quartz parameters are extracted first by sensor calibration. For three different quartz
resonators, the effective thickness and the static capacitance are similar. However, the effective
viscosities are different, which may be caused by different fixtures and surface toughness (Table
4).
Table 4 Quartz parameters by sensor calibration
Experiment number Effective viscosity Thickness (μm) Static capacitance (pF)
#1 0.0398 332 4.1352
#2 0.0239 332 4.57
#3 0.0378 332 4.41
24 hours after MC3T3-E1 cells cultured on the 5MHz TSM resonator surface, the
viscoelastic properties of the cell monolayer are extracted based on the methods discussed in
section 4.3. It is found that the values for Gcl', Gcl'' and hil obtained have wide ranges although
the theoretical values fit the measurements very well after each group of solution (Gclj', Gcl
j'' and
hilj) is plugged in the model (the error is less than 0.5%) (Figure 31). The wide ranges of Gcl',
86
Gcl'' and hil indicate that it is hard to extract three unknown parameters of loading layers at the
same time using the admittance spectra near resonant frequency. However, if the thickness of the
interfacial layer is known, the storage and loss shear modulus of the cell monolayer can be
obtained based on the relationships shown in Figure 32. Giebel et al. has reported that the values
of cell-substrate distance obtained with surface plasmon resonance microscopy are, respectively,
25nm for close contact and 160nm for the other part of cells [92]. It has also been reported by
Izzard and Lochner that the cell-substrate distance is about 30nm for the close contact, and is
100-140 nm for other regions of fibroblasts with interference reflection microscopy (IRM) [102].
And impedance measurements by cell-substrate electrical impedance sensing system showed that
the average cell-substrate distances were from 30-80 nm for fibroblasts [61]. In this study, we
assume the thickness of the interfacial layer between cell and substrate is in the range of 30-
80nm. The storage and loss shear modulus are therefore obtained. Three individual experiments
under the same conditions show similar results (Table 2).
Table 5 The extracted viscoelastic properties of cells for three individual experiments
Experiment number G' (KPa) G'' (KPa) η= G'' /ω(cp)
#1 21-30 49 1.56
#2 24-32 48 1.52
#3 24-37 29-35 0.92-1.11
87
0
0.0005
0.001
0.0015
0.002
0.0025
-0.001
-0.0005
0
0.0005
0.001
0.0015
0.002
4.999 5 5.001 5.002 5.003 5.004
G_actualG_calculated
B_actualB_calculated
G(s
) B(s)
Frequency (MHz)
Figure 31. Comparison of G (real part of the electrical admittance) and B (imaginary part of the electrical admittance) spectra for the resonators with cell adhesion between the actual experimental measurements and
calculated values by plugging the solutions into our theoretical model.
88
2 104
3 104
4 104
5 104
6 104
7 104
8 104
9 104
1 105
0 0.05 0.1 0.15 0.2
Experiment #2Experiment #1
Experiment #3
Stor
age
shea
r mod
ulus
(Pa)
Thickness of interfacial layer (μm)
(A)
0
1 104
2 104
3 104
4 104
5 104
6 104
7 104
0 0.05 0.1 0.15 0.2
Experiment #1Experiement #2Experimtent #3
Loss
she
ar m
odul
us (P
a)
Thickness of interfacial layer (μm)
(B)
Figure 32. The relationship between extracted values of the thickness of interfacial layer and storage shear modulus (A) and between extracted values of the thickness of interfacial layer and loss shear modulus (B).
89
Table 6 Comparison of cell storage and viscosity with different techniques
Method Cell type G'
(KPa)
Viscosity
(cp)
Measured
frequency
(Hz)
Model for
exacting cell
viscoelasticity
Ex-
ternal
Force
Re-
ference
TSM MC-3T3 21-38 0.92-1.56 5M No Our
study
Magnetic
Bead
Micro-
rheometry
NIH 3T3 20-40 4×106
Finite springs
and dashpots
model
Yes [65]
Magnetic
Bead
Micror-
heometry
Airway
smooth
muscle
cells
4 1.4×103 10
Power-law
structural
damping model
Yes [75]
Microplate Fibroblast
s 0.96 1.1×107
Finite springs
and dashpots
model
Yes [73]
AFM
Airway
smooth
muscle
cells
1.2 5.1×103 10
Power-law
structural
damping model
Yes [96]
90
Table 6 (continued)
Human
lung
epithelial
cells
1 1.68×103 10
power-law
structural
damping model
Yes [104]
AFM Chinese
hamster
ovary
(CHO)
cells
1.0-1.4 No [105]
Swiss
3T3
fibroblast
1.0-1.4 No
Mono-
nuclear
cell
1.0-1.5 No
Swiss
3T3 0.9-1.3 No
Micro-
fluori-
metry
MDCK 0.8-1.2 No
[106]
Our results are compared with those by other techniques, such as microplate method,
AFM, MBM and micro-fluorimetry (Table 6). Earlier techniques for characterizing cell
viscoelasticity include micropipette aspiration, magnetic cytometry and microplate manipulation,
91
in which the creep response and stress relaxation are measured and the simple viscoelastic
models with finite springs and dashpots are used to explain the experimental results [65, 73, 107].
More recently, dynamical viscoelastic properties of cells are characterized using magnetic
twisting cytometry and atomic force microscopy [75, 96, 104]. In these studies, the complex
shear modulus (G*(ω)) from oscillatory measurements over a frequency range is measured.
G*(ω) is defined as the complex ratio in the frequency domain between the applied stress and the
resulting strain. The functional dependence of modulus of cells is described by the power-law
structural damping model:
ωμωωηω
α
iiGG +⎟⎟⎠
⎞⎜⎜⎝
⎛+=
00 )1()(* (4.8)
where η is the hysteresivity or structural damping coefficient of the model, α is the power-law
exponent, and 0G and ω0 are scale factors for the moduli and frequency respectively [75, 96,
104]. This model shows that the storage modulus follows a power law with exponent α and the
loss modulus has the same power-law dependence on the frequency with an additional
Newtonian viscous term. At low frequencies, the loss modulus increases proportionally to G' (ω)
while at high frequencies G'' (ω) is dominated by the viscous term and approaches the linear
dependence of a Newtonian fluid.
Our measurements show the elastic shear modulus of cells at 5MHz is about 21-38KPa.
Dynamic measurements by AFM and MBM reveal the values for cell elastic modulus in the
range of 1-4 KPa at 10 Hz. According to Eq. (4.8), the elastic shear modulus has a power law
with exponent α, whose value is in the range of 0.15-0.35 according to previous experimental
results [75, 96, 104]. If we choose α=0.25 and Gcell'~30K at 5MHz in our case, we can calculate
Gcell'=1.1K at 10Hz, which is similar with the results by the dynamic measurement using AFM
92
and MBM at the same frequency range. However, the value by MBM and a simple viscoelastic
model with finite springs and dashpots in Bausch’s study is about 20-40KPa at quasi-static state,
about one order of magnitude larger than the dynamic measurements and our study [65]. Studies
by Desprat et al. have shown that viscoelastic properties extracted by different mathematic
models are quite different even though the experimental creep or stress relaxation curves are
obtained by the same microplate methods [108]. They conclude that the cells cannot be thought
as a viscoelastic medium with a few characteristic relaxation time constants as the finite springs
and dashpots models describing, but a viscoelastic medium with a wide and continuous spectrum
of relaxation time constants. The viscoelastic properties of cells obtained by finite springs and
dashpots model may not be reliable.
For the viscosity of cells, our results agree with those by micro-fluorimetry, which is a
non-invasive method to characterize viscosity of cell cytoplasm at very high frequencies. This
agreement is reasonable since cell viscosity in our study is measured at high frequency, which
has been shown relating to cytoplasm fluid. However, the value of Newtonian viscous term μ in
Eq. (4.8), indicating cell viscosity at high frequencies, by dynamic measurements with AFM and
magnetic bead microrhemethy is about three orders of magnitude larger than our results. Since
the dynamic measurements with AFM and MBM are implemented at relative low frequencies
and μ value shows cell viscosity at high frequencies, the method for calculating μ value by fitting
Eq. (4.8) to the experimental values of G*(ω) at relative low frequencies might cause errors. In
addition, the large probes in AFM and MBM may overestimate cell viscosity. Cell viscosity by
the static measurements and finite springs and dashpots models is even larger, about seven orders
larger than our results. Since the viscosity of cell is related to cytoskeleton at low frequencies but
related to cytoplasm fluid at high frequencies [109], the values of cell viscosity in our studies and
93
static measurements are related to the properties for different cellular structures. Another reason
is as we discussed, the finite springs and dashpots model might not be good enough to describe
cell viscoelasticity, making the results not reliable.
4.5 CONCLUSIONS
The present study has demonstrated that the TSM resonator can be effectively used as a
functional biosensor to determine viscoelastic properties of living cells. At the steady state of the
confluent cell layer adhering to the substrate, a theoretical model consisting of two uniform
layers (a finite uniform medium layer and a semi-infinite uniform viscoelastic cell layer) loaded
on the resonator surface is proposed to relate the electric response of resonators to mechanical
properties of cell layers. Using this model, the theoretical values of the resonator admittance near
resonance are fitted with the measured spectrum, and the viscoelastic properties of cell layers are
obtained. The results are reasonable compared with other techniques. In addition, the acoustic
wave resonator sensor has advantages of experimentally simple and non-invasive over other
techniques used to measure mechanical properties of cells.
However, in this study, the cell viscoelasticity cannot be obtained if the thickness of the
interfacial layer is unknown. The properties of interfacial layer are assumed to be the same as
those of cell culture medium, which may cause errors. In the future, the quartz TSM resonator
can be combined with electrical impedance sensing system, which can measure cell-substrate
separation, to characterize the cell viscoelasticity. The properties of interfacial layer could be
characterized by measuring the electrical response of TSM resonators at higher order resonant
frequencies so that the acoustic wave only penetrates into the interfacial layer.
94
5.0 LOVE MODE SURFACE ACOUSTIC WAVE DEVICE AS CELL-BASED
BIOSENSOR
5.1 INTRODUCTION
In chapters 3 and 4, we have discussed the application of thickness shear mode bulk acoustic
wave sensors for monitoring cell adhesion and characterizing cell viscoelasticity. Is it possible to
use the surface acoustic wave devices as cell-based sensors for cell biological studies? What are
the advantages and disadvantages of surface acoustic wave devices as cell-based biosensors
compared with thickness shear mode bulk acoustic wave devices? Of various types of surface
acoustic wave devices, wave-guided shear-horizontal surface acoustic wave device have
attracted more and more attention due to their advantages over other acoustic wave devices for
sensing applications in liquid, such as high sensitivity, mechanical robustness, and the protection
of the IDTs due to the use of the waveguide layer. As talked in chapter 1, wave-guided shear-
horizontal surface acoustic wave (also called Love mode) devices use piezoelectric substrate
with inter-digital transducers (IDTs) deposited on one side of the surfaces to produce the shear-
horizontal acoustic waves. In order to prevent the energy lost into the bulk material, a material
layer with a slower acoustic velocity as a waveguide are deposited on it to trap the acoustic
energy near the surface (figure 12) [3, 31]. However, the current investigation on Love mode
biosensors is still far from mature, for both the theoretical and experimental studies. Most of
95
studies are limited to the use of Love mode devices as mass sensors. No studies have been
performed to probe cell behaviors and properties using Love mode devices so far. In this chapter,
we developed a theoretical model to optimize the sensor performance and predict the electrical
signals changes (e.g. resonant frequency and attenuation) due to cell adhesion to the sensor
surface. Experimentally, we fabricated Love mode acoustic wave device and experimentally
demonstrated the validity of our theoretical model.
5.2 THEORETICAL MODELING OF CELL-BASED LOVE MODE ACOUSTIC
WAVE SENSORS
As discussed in chapter 1, up to now, there are several methods developed to theoretically
describe Love mode acoustic wave sensors, including analytical method [26, 27], perturbation
method, [3, 28, 29], and transmission line method [30]. These methods have their limitations. For
example, the analytical method is computationally complex and difficult to relate the changes in
the loading to corresponding changes in electrical signals of the device, especially in the case of
considering the loss of materials. Perturbation method assumes the changes in the system
parameters are very small. Therefore, the perturbation method is limited to the modeling of the
small changes of physical parameters in the system. The transmission line model has been
applied for the modeling of Love mode acoustic wave devices as sensors [31]. It has been
demonstrated that this model can describe the behavior of the sensor loaded by viscoelastic
masses in a viscous liquid. However, the calculation is still complex. Therefore, it is necessary to
develop an accurate and efficient method to describe the Love mode acoustic wave sensors with
various loading, especially multi-viscoelastic layer loading. The transfer-matrix method is used
96
when the total system can be broken into a sequence of subsystems that interact only with
adjacent subsystems. In our previous studies, it has been successfully used in the modeling of
quartz thickness shear mode resonators as cell-based sensors [110]. It is a simple method for
modeling the acoustic wave devices, especially for the multilayer structures. So the transfer
matrix method is suitable for the analysis of Love mode devices as cell-based biosensors.
Piezoelectric substrate
Wave-guiding layerLoading layer 1Loading layer 2Loading layer 3
Loading layer n
v3
F3
v2
F2
vL
FL
B1 B2substrate Vn+1
Fn+1=0
vn
Fn
Bn
x3
x1
x3
x2
A
B
Figure 33. (A) Models for Love mode acoustic wave sensors with multilayer loadings; (B) Transfer matrix model of a Love mode acoustic wave device with multiple uniform layers loaded on one side
97
For Love mode acoustic wave sensors with multilayer loadings, the sensor structure
includes a semi-infinite piezoelectric substrate, a finite wave-guiding layer and multiple loading
layers (Figure 33A). According to the transfer matrix model (Figure 33B), the transfer matrix B
for non-piezoelectric uniform layer is given by
[ ] ⎥⎦
⎤⎢⎣
⎡×⎥
⎦
⎤⎢⎣
⎡=⎥
⎦
⎤⎢⎣
⎡×=⎥
⎦
⎤⎢⎣
⎡
i
i
i
i
vF
ZjjZ
vF
BvF
γγγγ
cossinsincos
0
0 (5.1)
where F and v represent stress and particle velocity, respectively; subscript i and o represent
input and output ports of the layer; Z and γ are the characteristic acoustic impedance and the
phase delay of the layer [87]. Since the direction of transfer matrix model is direction, the
characteristic acoustic impedance Z and phase delay are calculated in direction. In the Love
mode devices, we assume that the acoustic waves are pure shear-horizontal. Therefore, the wave
equation in the layers can be written as
3x
3x
)(exp()())()((exp()(),,( 3233311132312 rKtjxUxjkxjktjxUtxxU r−=−−−−= ωααω (5.2)
where U is the particle displacement, ω is the angular frequency, k is wave number, α is
attenuation (rate of amplitude change with the propagation distance), t is time, and subscriptions
1, 2, and 3 represent x1, x2 and x3 directions respectively. The complex wave number of the
acoustic wave (2/12/1233
211 )/())()(( GjkjkK ρωαα =−+−= ρ and G are the density and
complex shear modulus of the layer), and the propagation velocity of the acoustic wave in the
device 1/ kv p ω= . The characteristic impedance Z and phase delay in the layer γ in direction
are given by
3x
98
G
jk
GjkGGKGZ ))(
())()/((2
11
2/1211
23
αω
ραρωωω
−
−=−−== (5.3)
G
jk
GGhjkGhhKh )
)(())()/((/2
2
11
2/1211
23
αω
ρωαρωλπγ
−
−=−−=== (5.4)
where K3= 33 αjk − , which is complex wave number of the acoustic wave in direction, and h
is the thickness of the layer. Comparing the characteristic acoustic impedance and the phase
delay for the thickness shear mode resonator, which are given by
3x
GZ ρ= and Gh /ρωγ = ,
with those for Love mode devices, we found that the expressions of Z and γ in these two modes
were similar. The differences bwteen them are caused by the different propagation directions of
acoustic waves: in the Love mode acoustic wave devices, the acoustic wave propagates in both
and directions, while in the thickness shear mode resonators, the wave propagates only in
direction. According to Eqs. (5.1), (5.3) and (5.4), the transfer matrix of the finite non-
piezoelectric layer is given by
3x 1x
3x
[ ]
⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
⎟⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜⎜
⎝
⎛
−
−
−
−
⎟⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜⎜
⎝
⎛
−
−
⎟⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜⎜
⎝
⎛
−
−
−
−
⎟⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜⎜
⎝
⎛
−
−
=
G
jk
GGhG
jk
GG
jk
GGhj
G
jk
GGhG
jk
GjG
jk
GGh
B
))(
(cos))(
())(
(sin
))(
(sin))(
())(
(cos
2
11
2
11
2
11
2
11
2
11
2
11
αω
ρω
αω
ρ
αω
ρω
αω
ρω
αω
ρ
αω
ρω (5.5)
With these transfer matrixes, the overall acoustic impedance of uniform multilayer loading ZL
can be easily calculated using the following equation:
[ ] [ ] [ ] ⎥⎦
⎤⎢⎣
⎡××⋅⋅⋅××=⎥
⎦
⎤⎢⎣
⎡
+
+
1
121
n
nn
L
L
vF
BBBvF
(5.6)
where Fn+1=0, since the top layer is exposed to air.
99
In chapter 3, we have developed a multiple layer model for cells adhering onto the
thickness shear mode resonators. Similarly, the cell-based Love mode acoustic wave biosensors
can be considered to include four uniform layers, which are the semi-infinite piezoelectric
substrate, finite wave guiding layer, finite interfacial layer and semi-infinite cell layer. Therefore,
according to Eq. (5.6), the overall acoustic loading impedance applied onto the piezoelectric
substrate when cells adhere to the substrate is given by
)
))(
(
tan())(
(
)
))(
(
tan())(
(
))(
(
2
111
2
11
2
11
2
11
1
2
11
waveguidwaveguid
waveguidwaveguid
Lwaveguidwaveguid
waveguid
waveguidwaveguid
waveguidwaveguid
waveguidwaveguid
waveguidL
waveguidwaveguid
waveguidL
hG
jk
G
jZG
jk
G
hG
jk
G
G
jk
GjZ
G
jk
GZ
αω
ρ
ω
αω
ρ
αω
ρ
ω
αω
ρ
αω
ρ
−
−
+
−
−
−
−
−
−+
−
−=
(5.7)
where subscription waveguide represents wave guiding layer and ZL1 is the acoustic loading
impedance on the wave guiding layer, which is calculated by
)
))(
(
tan())(
())(
(
)
))(
(
tan())(
())(
(
))(
(
2
11
2
11
2
11
2
11
2
11
2
11
2
11
1
ilil
ilil
clcl
clilil
il
ilil
ilil
ilil
ilclcl
cl
ilil
ilL
hG
jk
G
G
jk
GjG
jk
G
hG
jk
G
G
jk
GjG
jk
G
G
jk
GZ
αωρ
ω
αωρ
αωρ
αωρ
ω
αωρ
αωρ
αωρ
−
−
−
−+
−
−
−
−
−
−+
−
−
−
−=
where subscription il and cl represent the interfacial layer and cell layer respectively. In the Love
mode acoustic wave sensors, the wave equations in the semi-infinite piezoelectric substrate is
given by
)))((exp())exp(( 11132 xjktjxBjAuu oss αω −−⋅+= (5.8)
100
Although the piezoelectric substrate used here is anisotropic material, for the simplicity, in the
calculation, first, we treated the substrate as an isotropic material. Therefore, the equation of
motion in piezoelectric substrate can be written as
0))(()2( 211
222 =−−+++− sss GjkBjABAG αωρ (5.9a)
⎪⎩
⎪⎨⎧
−−=−⇒=+−+−
−=⇒=+⇒ 2
12
12
21222
12
1222
1111
))/(
1(0)(
0)(
αω
αωρ
αα
kv
kABGGkABG
kABkABjG
sssss
s
(5.9b)
where s
ss
Cvρ
=2 . Subscription s represents the piezoelectric substrate and ρ is the density. From
the substrate direction, the acoustic loading impedance is given by
0
0
2
3
2
02
320
3
3
3
3
)(
=
=
==
+=
∂∂
==x
s
x
sx
xs jBjAG
vxu
GvTZ
ω (5. 10)
which is equal to -ZL. So the solution of the wavenumber k1 and α are obtained. Finally, the
phase velocity and attenuation of the shear-horizontal acoustic waves in Love mode sensors can
be calculated according to the values of k1 and α. The transfer matrix method provides a simple
way to evaluate the impact of the modification at the surface of the sensor. Indeed, we can
calculate the basic properties of the structure on the devices, such as a rigid or viscoelastic layer
and with or without a liquid environment.
101
5.3 OPTIMIZATION OF LOVE MODE ACOUSTIC WAVE SENSOR DESIGN
It has been noted that several parameters of Love mode devices, such as operating frequency,
thickness and mechanical properties of wave-guiding layers, affect the sensing performance of
the device [26, 31, 111]. In this section, we optimize the design of the acoustic structure and
selection of materials that enhance the sensitivity according to our theoretical model developed
in last section. To optimize the sensor design, first we have to investigate the effects of design
parameters (material properties and thickness of wave-guiding layer, material properties of the
piezoelectric substrate) on the sensor performance (sensitivity). The effect of the properties of
the guiding layer and piezoelectric substrate on the acoustic wave propagation velocity was
simulated using our theoretical model in the last section first. In this simulation, we assume that
all the materials for wave guiding layers are acoustic lossless. From the modeling results, we
found that the changes in propagation velocity with the increase in the thickness of wave guiding
layer were periodic. Τhe period decreases with the decrease in the phase velocity of the guiding
layer, while increases with the decrease in the phase velocity of the piezoelectric substrate
(Figure 34A). However, the density of piezoelectric substrate and guiding layer don’t affect the
period (Figure 34B). In addition, we found that a significant negative shift in the propagation
velocity occurred when the phase shift ϕ3 in the wave guiding layer in the thickness direction is
around odd times of π/2 (Figure 34C). So the period of propagation velocity change is
associated with phase shift in the guiding layer. Rewriting Eq. (5.4) as 22
11vv
hgl
−= ωγ , we get the
period is given by )11/( 22 vvgl
−ωπ , which confirms that smaller vgl results in a smaller period
while the density of the substrate and guiding layer does nothing with it. In our simulation, the
102
operating frequency is calculated by substratev*)/2( λπ , in which λ is a constant. So the larger
phase velocity causes the larger operating frequency, which confirms our simulation results too.
0
2000
4000
6000
8000
0 1 10-6 2 10-6 3 10-6 4 10-6 5 10-6 6 10-6 7 10-6
vs=5060m/s; v
gl=707m/s
vs=5060m/s; v
gl=1000m/s
vs=5060m/s; v
gl=1414m/s
vs=4108m/s; v
gl=1000m/s
Prap
agat
ion
velo
city
(m/s
)
Thickness of guiding layer (m)
A densitys=2648
densitygl=1000
1000
2000
3000
4000
5000
6000
0 1 10-6 2 10-6 3 10-6 4 10-6 5 10-6 6 10-6
densitys=2648; density
gl=1000
densitys=2648; density
gl=2000
densitys=2648; density
gl=3000
densitys=7456; density
gl=1000
Pro
paga
tion
velo
city
(m/s
)
Thickness of wave guiding layer (m)
B
vs=5060m/s
vgl=1000m/s
103
0
2
4
6
8
10
12
14
0 2 10-6 4 10-6 6 10-6 8 10-6 1 10-5
vs=5060m /s; v
g l=707m /s; densi ty
s=2648; densi ty
gl=1000
vs=5060m /s; v
g l=1000m/s; density
s=2648; density
g l=1000
vs=5060m /s; v
g l=1414m/s; density
s=2648; density
g l=1000
vs=5060m /s; v
g l=1000m/s
densitys=2648; density
g l=2000
vs=5060m /s; v
g l=1000m/s
densitys=2648; density
g l=3000
Phas
e sh
ift in
wav
e gu
idin
g la
yer (
rad)
Thickness of wave guiding layer (m)
C
0
5 106
1 107
1.5 107
2 107
2.5 107
-0.5
0
0.5
1
1.5
2
2.5
3
3.5
-5 10-7 0 5 10-7 1 10-6 1.5 10-6 2 10-6 2.5 10-6 3 10-6 3.5 10-6
s1
s2
s3
s4
ϕ1
ϕ2
ϕ3
ϕ4
Mas
s se
nsiti
vity
(m3 /s
*kg)
Phase shift in w
ave guiding layer (rad)
Thickness of wave guiding layer (m)
D
Figure 34. (A) Propagation velocity v.s. the thickness of the guiding layer; (B) Propagation velocity v.s. the thickness of the guiding layer; (C) Phase shift in wave guiding layer v.s. the thickness of wave guiding layer; (D) mass sensitivity and phase shift in guiding layer v.s. the thickness of guiding layer. The materials parameters of
substrate and guiding layer for each curve are (vs=5060m/s, vgl=707m/s, densitys=2648kg/m3, densitygl=1000kg/m3 for s1 and ϕ1; vs=5060m/s, vgl=1000m/s, densitys=2648kg/m3, densitygl=1000kg/m3 for s2 and ϕ2; vs=5060m/s,
vgl=1414m/s, densitys=2648kg/m3, densitygl=1000kg/m3 for s3 and ϕ3; vs=4108m/s, vgl=1000m/s, densitys=2648kg/m3, densitygl=1000kg/m3 for s4 and ϕ4)
104
For the mass sensitivity, we have found that it reaches a maximum value when the phase
shift in the guiding layer is around π/2 for each device. The maximum mass sensitivity is larger
for the devices with larger vsubstrate and smaller vgl (Figure 34D). The value of π/2 for the phase
shift in the guiding layer indicates that the film resonance happens at that time and the shear
stresses applied to the lower film surface interfere constructively with those reflected from the
upper film surface. At that condition, the maximum coupling of acoustic energy from the
piezoelectric substrate to the wave guiding layer is obtained, which causes the highest sensitivity.
In addition, the larger vsubstrate and smaller vgl results in the deposition of a thinner guiding layer
to obtain the film resonance, which means in the devices with the larger vsubstrate and smaller vgl,
the acoustic energy is trapped in a thinner layer. This results in the increase in the device
sensitivity.
Considering the guiding layer is acoustic loss, we checked the effect of the guiding layer
viscosity on the propagation velocity, attenuation and their sensitivity to mass loading. The
guiding layer with acoustic loss not only induces changes in phase velocity, but also attenuation
in acoustic waves. With increasing the viscosity of guiding layer, the changes in the propagation
velocity and attenuation are significantly damped, especially when the guiding layers are at the
resonance in the thickness mode. In addition, compared with the fundamental resonant response,
the harmonic resonant responses for both propagation velocity and attenuation become less
pronounced (Figure 35A and 35B). It has been explained that particle motion at the upper
surface of the film goes from an in-phase to an out-of-phase condition when the phase shift is
beyond π/2, which results in a smaller particle excursion in the upper regions of the film,
diminishing the kinetic contribution to velocity and attenuation response [20]. Figure 35C shows
the response of propagation velocity and attenuation sensitivity to mass loading with changing
105
the viscosity and the thickness of guiding layer. We found that with increasing the viscosity of
guiding layer, the maximum values of the propagation velocity and attenuation sensitivity to
mass loading (for different thickness of guiding layer) increased first and then decreased when
the viscosity of guiding layer is around 80cp and 160cp respectively. In addition, we found the
sensitivity was affected by the original acoustic loading on the device. In this simulation, we
calculated the sensitivity by computing the velocity and attenuation changes induced by the gold
layer with 1nm in thickness when 0.1nm, 1nm, and 10nm gold layer already deposited onto the
devices. We found the original acoustic loading also affects the sensitivity, especially for the
thickness of guiding layer on which the sensitivity has the maximum value.
106
2000
3000
4000
5000
6000
0 1 2 3 4 5 6 7 8
ηgl=0
ηgl=0.08
ηgl=0.16
ηgl=0.2
Pro
paga
tion
ve
loci
ty (m
/s)
Thickness of wave guid ing layer (μm)
A
-4 104
-2 104
0
2 104
4 104
0 1 2 3 4 5 6 7 8
ηgl=0
ηgl=0.08
ηgl=0.16
ηgl=0.2
Att
enu
atio
n α
Thickness of wave guiding layer (μm)
B
107
-2 106
0
2 106
4 106
6 106
8 106
1 107
1.2 107
1.4 107
0.6 0.8 1 1.2 1.4 1.6 1.8 2
ηgl=0,h
gl=1e-10
ηgl=0,h
gl=1e-9
ηgl=0,h
gl=1e-8
ηgl=0.08,h
gl=1e-10
ηgl=0.08,h
gl=1e-9
ηgl=0.08,h
gl=1e-8
ηgl=0.16,h
gl=1e-10
ηgl=0.16,h
gl=1e-9
ηgl=0.16,h
gl=1e-8
Pro
paga
tion
vel
ocity
sen
sitv
ity to
mas
s lo
adin
g (m
3 /(kg*
s))
Thickness of wave guiding layer (μm)
C
0
400
800
1200
1600
0.6 0. 8 1 1.2 1.4 1.6 1.8 2
ηgl
=0,hgl
=1e -10
ηgl
=0,hgl
=1e -9
ηgl
=0,hgl
=1e -8
ηgl=0.08, hgl=1e -10
ηgl=0.08, hgl=1e -9
ηgl
=0.08, hgl=1e -8
ηgl
=0.16,hgl
=1e -10
ηgl
=0.16,hgl
=1e -9
ηgl
=0.16,hgl
=1e -8
ηgl=0.2,hgl=1e -10
ηgl=0.2,hgl=1e -9
ηgl
=0.2,hgl
=1e -8
Att
enua
tion
sen
sitv
ity t
o m
ass
load
ing
(rad
*m2 /k
g)
Th icknes s o f wa ve g uidin g laye r (μ m)
D
Figure 35. (A) Propagation velocity v.s. thickness of wave guiding layer; (B) Wave attenuation v.s. thickness of wave guiding layer; (C) Propagation velocity sensitivity to mass loading v.s. thickness of wave guiding layer; (D)
Attenuation sensitivity to mass loading v.s. thickness of wave guiding layer
108
Table 7 Parameters of the materials for the simulation of Love mode devices
Material Density (kg/m3)Phase velocity
(m/s)
Shear stiffness
(GPa) Viscosity (cp)
Quartz 2648 5060 66.3 0
36o rotated YX
LiTaO3
7456.4 4108 130.3 0
PMMA 1100 1100 1.3 14.14
SiO2 2610 2850 8.3 0
SU-8 1200 814 0.796 1.414*10-2
Based on our simulation results, we conclude that in order to obtain a high sensitivity for
the Love mode device, the wave guiding layer and piezoelectric substrates should meet several
requirements: high phase velocity for the piezoelectric substrate and low phase velocity and
relative low viscosity (less than 80-100cp) for the wave-guiding layer. In addition, we should
control the thickness of guiding layer to make the phase shift in the layer around π/2. Table 7
lists the parameters of the materials usually considered as the substrate and wave guiding layer
for Love mode devices. ST-cut quartz and 36o rotated YX LiTaO3 are usually used as the
substrate, and PMMA, SiO2 and SU-8 as the guiding layers. The phase velocity of 36o rotated
YX LiTaO3 is a little lower than that ST-cut quartz. However, since 36o rotated YX LiTaO3 has a
relatively higher electromechanical coupling coefficient than quartz, which makes the better
performance of the device. So both quartz and YX LiTaO3 can be considered as the piezoelectric
substrate. For the wave guiding layer, PMMA and SU-8 have the similar properties, lower phase
velocity but higher viscosity compared with SiO2. Since the viscosity for PMMA and SU-8 is
109
much less than 80cp, which does not induce a significant decrease in the sensitivity according to
our simulation results, we will choose PMMA as the wave-guiding layer in our experimental
studies.
5.4 EXPERIMENTAL CHARACTERIZATION OF LOVE MODE ACOUSTIC
WAVE SENSORS
5.4.1 Fabrication of Love mode acoustic wave devices
The piezoelectric substrate used for the Love mode acoustic wave device is a 36o-rotated Y-cut
X-propagation Lithium Tantalate (36oYX-LiYaO3). The interdigital transducers (IDTs) were
fabricated using the photolithography and lift-off procedures. Basically, photoresists were
patterned on the Lithium Tantalate using standard photolithography techniques. The 10/80nm
thick Cr/Au layer was deposited onto the photoresist-patterned substrate, covering the whole
surface. During the lifting-off, the photoresist under the metal film is removed with solvent,
taking the film with it, and leaving only the film which was deposited directly on the substrate.
After finishing patterning IDTs on the substrate, the wafer is diced into several devices. Each
device has a dual delay line configuration, of which one is the reference line and the other is
sensing line. This dual line is used to reduce all the secondary environmental effects such as
temperature variation, pressure interaction, etc (Figure 36). To reduce the reflection of acoustic
waves from the crystal ends and bulk wave component that can affect the sensitivity and
reproducibility of the device, the two short ends of the substrate, which are perpendicular to the
direction of wave propagation, were coated with PDMS as a absorber. The PDMS can absorb the
110
waves to avoid the reflection of the acoustic wave from the two short ends of the substrate. The
bottom side of the device is roughed to dissipate bulk acoustic wave that are generated by the
transducers. For each IDTs, there are thirty split finger pairs IDTs having periodicity of 32μm.
Therefore, the operating frequency is around 131MHz. The IDT center-to-center separation is
about 8.8 mm and the thickness of the substrate is about 0.5mm.
Figure 36. Fabricated shear-horizontal acoustic wave devices
PMMA was used as the wave-guide layer to trap the acoustic energy near the sensing
surface in our experiments. According to the thickness range of the PMMA film desired, PMMA
solution with appropriate molecular weight and concentration was chosen. Basically, PMMA
was coated onto the devices by spin coating. To characterize our device, we coated the PMMA
film with the thickness of 1.4-3.3 μm by spin coating with appropriate spin speed and time. The
device with PMMA coating was then cured at 180oC for 2 hours in ovens. After curing, the
111
coated devices were allowed to cool down to room temperature. The thickness of PMMA films
was measured using the surface profilometer then. After the PMMA film coated on the device
surface, platinum wires were bounded to the electrode pads using conductive silver paste (SPI
Supplies and Structure Probe, Inc., West Chester, PA). Then the devices were baked at 120oC on
the hot plate for several minutes. The uncured PDMS were poured at the two short ends of the
devices. At high temperature, PDMS was quickly cured and severs as absorbers. Finally, the
Love mode acoustic wave devices were obtained (Figure 37).
Figure 37. Love mode acoustic wave devices
112
5.4.2 Experimental setup
During the measurement, the acoustic wave sensors were mounted onto the fixtures by tapes
(Figure 38).
Figure 38. Love mode acoustic wave devices mounted on the test fixture
The wires glued to IDT electrodes were linked to BNC connectors in the test fixture,
which were connected to the measurement system. The measurement system is comprised of a
vector network analyzer (HP8722C), a switch/control unit (Agilent 34970A), hi-speed GPIB
controller and the labview software (Figure 39).
113
Love mode sensors
Switch Unit
Network Analyzer
Hi-Speed GPIB Controller
Labview Program
Love mode sensors
Switch Unit
Network Analyzer
Hi-Speed GPIB Controller
Labview Program
Figure 39. Measurement system of Love mode acoustic wave sensors.
5.4.3 Experimental results
First, the devices without PMMA coatings were mounted onto the test fixture. The amplitude and
phase of parameter S21 between two IDTs were measured using the network analyzer. It was
found that the resonant frequency of all the devices was around 130MHz and the insertion loss at
the resonant frequency was about 29dB. The relative high insertion loss is because IDTs on the
36o rotated LiTaO3 substrate produces the surface skimming bulk wave, in which the acoustic
waves are propagating deeper into the bulk material therefore the insertion loss increases. After
the measurement of the devices without PMMA coating, PMMA were spun on the device
surface with different spinning speed and time in order to obtain the different film thickness in
the range of 1,2-3.5μm. Table 8 shows the experimental conditions and thickness of the PMMA
film under each condition.
114
Table 8 Experimental conditions of PMMA spin coating and the thickness of PMMA film
Material spinning speed (rpm) spinning time (s) Film thickness (μm)
PMMA 950 C7 4000 30 1.0
PMMA 950 C7 2500 30 1.4
PMMA 950 C9 4000 30 2.0
PMMA 950 C7 1500 30 2.2
PMMA 950 C9 3500 30 2.4
PMMA 950 C9 3000 30 2.7
PMMA 950 C9 2500 30 2.9
PMMA 950 C9 2000 30 3.3
PMMA 950 C9 1500 30 3.5
The S21 between two IDTs on the devices with PMMA coatings with different thickness
was measured then. The amplitude and phase images of the device with 2.0μm thick PMMA
coating showed that the resonant frequency decreases a lot, which is about 6MHz, compared
with uncoated device while the insertion loss decreased since the PMMA layer traps the acoustic
energy in the guiding layer (Figure 40).
115
Figure 40. The amplitude and phase images of S21for Love mode acoustic sensors with 2.0mm PMMA coating.
With the increase in the thickness of the PMMA film, the resonant frequency decreases a
little bit first and decreases dramatically when the thickness of PMMA film is around 2.0μm.
When the thickness of PMMA layer is larger than 2.5μm, the resonant frequency returns to 130
MHz again. Comparing our experimental results with our theoretical modeling ones, we found
that the trend of resonant frequency change with increasing the thickness of PMMA coating layer
116
was consistent. However, the maximum frequency shift obtained by our experimental results is
much smaller than that predicted by the theoretical models (Figure 41). The reason is that
according to our theoretical models, the significant decrease in the resonant frequency occurs
when the thickness of PMMA coating layer is in the range of 2.1-2.3μm, which is a very narrow
range. In our experiment, we used spinning coating method to deposit PMMA layer on the
sensor surface, which is very hard to control the thickness of the coating layer precisely.
Therefore, it is hard to obtain the PMMA film with a certain thickness that causes a large
decrease in the resonant frequency of the Love mode acoustic wave sensors.
The initial insertion loss for the device without PMMA coating is about -29dB.
Compared with bared device, the insertion loss change induced by the PMMA coating layer
initially increases with increasing the PMMA guiding layer thickness, reaching a minimum value
at about 1.4μm. Then the increase in the thickness of the PMMA layer gives a gradual decrease
in insertion loss (Figure 42). Comparing the experimental results with our theoretical modeling
results, we found that the increase in the insertion loss could not be predicted by our theoretical
models. In addition, our theoretical model showed that the insertion loss for the device without
coating was zero while our experimental results showed 29dB attenuation from the input IDT to
output IDT. Although our theoretical model cannot predict the initial signal attenuation for the
bared device and the increase in the insertion loss when the thickness of PMMA guiding layer is
thin, the gradual decrease after the insertion loss reaches a maximum value observed in our
experimental results, which is consistent with our theoretical modeling results. The decrease in
the insertion loss with increasing the thickness of PMMA guiding layer is believed to be
determined by the viscosity of PMMA wave guiding layer.
117
The reason that causes the inconsistence between our theoretical and experimental results
is that in our theoretical modeling, we assume all the acoustic energy is located near the surface
of shear-horizontal acoustic wave device and the output IDT can transduce all the acoustic
energy in the device into electrical signals. Therefore, the insertion loss in our theoretical model
is assumed to be the attenuation in the acoustic waves when they propagate in x1 direction. So in
our calculation, we use parameter α to compute the attenuation between two IDTs. However, for
the real device, as we discussed in chapter 1, there is no pure shear-horizontal surface acoustic
wave device. The IDTs on the 36o rotated LiTaO3 substrate produces the surface skimming bulk
wave, in which the acoustic waves are propagating deeper into the bulk material. Therefore, the
output IDT cannot transform all the acoustic energy into the electrical signals and the additional
signal attenuation between two IDTs is produced. Therefore, in real experimental measurements,
the insertion loss is the sum of the loss induced by material viscosity when the acoustic wave
propagates in it in x1 direction, the energy loss due to the acoustic wave propagates into the bulk
material, and the loss induced by measurement system. In our theoretical modeling, we only
considered the loss induced by the loss of wave-guiding materials, which causes the inconsistent
between our theoretical modeling results and experimental results.
Overall, our experimental results demonstrated that our theoretical model could
efficiently predict the resonant frequency shift in the Love mode acoustic wave device with
loadings. For the insertion loss, although our models cannot describe the signal attenuation
induced by the changes in acoustic energy distribution because of the wave guiding layer
deposition, they are able to predict the insertion loss induced by the material loss in loading
layers. For the biological applications of Love mode acoustic wave devices, appropriate wave
guiding layers have been chosen and deposited on the device surface. Therefore, most of energy
118
has been trapped near the device surface. With cells attaching to the surfaceof this device, most
of the acoustic energy is still confined in or near the guiding layers. Therefore, the insertion loss
due to cells is mainly induced by the material loss of the cell, protein or medium layers. So our
theoretical models are valid for describing the resonant frequency and insertion loss change for
Love mode acoustic wave devices as cell-based biosensors.
Thickness of PMMA coating layer (m)
0.0 4.0e-7 8.0e-7 1.2e-6 1.6e-6 2.0e-6 2.4e-6 2.8e-6 3.2e-6
Res
onan
t fre
quen
cy (H
z)
8.0e+7
1.0e+8
1.2e+8
1.4e+8
1.6e+8
Modeling result (viscosity_PMMA=14cp)Modeling result (viscosity_PMMA=60cp)Experimental result
Figure 41. Comparison of theoretical and experimental results of resonant frequency shift with the thickness of PMMA coating layer.
119
Thickness of PMMA coating layer (m)
0.0 4.0e-7 8.0e-7 1.2e-6 1.6e-6 2.0e-6 2.4e-6 2.8e-6 3.2e-6
Inse
rtio
n lo
ss c
hang
e(dB
)
-80
-60
-40
-20
0
Modeling results (η_PMMA=1.4cp)Modeling results (η_PMMA=14cp)Modeling results (η_PMMA=140cp)Experimental results
Figure 42. Comparison of theoretical and experimental results of insertion loss changes induced by PMMA layer with the different thickness.
5.5 POTENTIAL APPLICATIONS OF LOVE MODE ACOUSTIC WAVE SENSOR
FOR CELL BIOLOGICAL STUDIES
In the above sections, we have discussed the theoretical modeling, the optimization, and the
experimental characterization of the Love mode acoustic wave sensors. Our experimental results
showed the ability of our theoretical model to help optimize the sensor design and its validity. In
this section, through investigating the relationship between the electrical signals of the device
120
(resonant frequency and attenuation) and cell-substrate interaction and cell viscoelastic
properties, we discuss the potential applications of Love mode acoustic wave sensors for cell
biological studies.
The effect of the thickness of the interfacial layer and wave guiding layer on the relative
changes in the propagation velocity and attenuation to the semi-infinite cell medium loading was
investigated (Figure 43). In this simulation, we choose 36o rotated LiTaO3 as substrate and
PMMA as wave guiding layer. According to our previous study, the storage shear modulus and
viscosity of cell layer are chosen to be 50KPa and 1cp respectively. The viscosity of interfacial
layer is assumed to be 2cp [73, 75, 104, 105]. The relative change in phase velocity and
attenuation to the semi-infinite cell medium loading is caused by the cell layer. From the results,
we can see that when the thickness of the interfacial layer is beyond about 150nm, the relative
change in phase velocity and attenuation to the semi-infinite cell medium loading almost reaches
to zero. These results indicate that the penetration depth of the acoustic wave with the frequency
of 131MHz (the operating frequency in our simulation) in the medium with the viscosity of 2cp
is about 150nm. In another word, if the cell layer is too far from the substrate, for the Love mode
acoustic wave devices with the operating frequency above 100MHz, the acoustic waves are
totally damped before they reach the cell layers. Therefore, for Love mode acoustic wave devices
with operating frequency above 100MHz, cells can be detected only when cells tightly attach to
the substrate. If the cells are far from the substrate, the acoustic waves only penetrate into the
interfacial layer, therefore, the devices has the potential to detect the properties of interfacial
layer, which may be altered by the proteins in this layer and focal adhesion area between the
substrate and cell layer. Therefore, the Love mode acoustic devices have the potential to detect
the protein production by cells and monitor cell adhesion to the substrate.
121
-20
-15
-10
-5
0
5
10
15
0 0.0 5 0.1 0 .15 0.2 0.2 5 0.3 0.35
hgl=1.4μ m
hgl=1.6μ m
hgl=1.7μ m
hgl=1.8μ m
hgl=1.9μ m
hgl=2.0μ m
Pro
pag
atio
n v
elo
city
cha
nge
(m
/s)
Thic knes s of interfacia l lay er (μ m)
A
Gcl'=50 KPa
hcl
=1cp
hil=2cp
-2500
-2000
-1500
-1000
-500
0
500
0 0.05 0.1 0.15 0.2 0.25
hgl=1.4μm
hgl=1.6μm
hgl=1.7μm
hgl=1.8μm
hgl=1.9μm
hgl=2.0μm
Att
enua
tion
α c
hang
e
Thickness of interfacial layer (μm)
Gcl'=50KPa
hcl=1cp
hil=2cp
B
Figure 43. Effect of the thickness of interfacial layer and wave guiding layer on the relative changes in (A) the propagation velocity and (B) attenuation to the semi-infinite cell medium loading.
122
-3
-2.5
-2
-1.5
-1
-100
-90
-80
-70
-60
-50
-40
-30
-20
104 105
Velocity
Attenuation
Prop
agat
ion
velo
city
cha
nge
(m/s
)A
ttenuation change α (rad)
Storage shear modulus of cell layer (KPa)
hgl'=1.75μm
ηcl=1cp
ηil=2cp
hil=50nm
A
-4
-2
0
2
4
-100
-50
0
50
100
0.001 0.002 0.003 0.004 0.005 0.006 0.007
velocity
attenuation
Prop
agat
ion
velo
city
cha
nge
(m/s
)A
ttenuation change α (rad)
Viscosity of cell layer (1000cp)
hgl=1.75μm
Gcl'=50KPa
ηil=2cp
hil=50nm
B
123
-25
-20
-15
-10
-5
0
5
10
15
-300
-200
-100
0
100
200
0 0.002 0.004 0.006 0.008 0.01 0.012
velocity
attenuation
Prop
agat
ion
velo
city
cha
nge
(m/s
)A
ttenuation change α (rad)
Viscosity of interfacial layer (1000cp)
hgl=1.75μm
Gcl'=50KPa
ηcl=1cp
hi l=50nm
C
Figure 44 Effects of (A) storage shear modulus of cell layer, (B) viscosity of cell layer and (C) viscosity of interfacial layer on the relative changes in the propagation velocity and attenuation to the semi-infinite cell medium
loading.
The effect of the storage shear modulus of cell layer on the propagation velocity and
attenuation of Love waves was investigated using our theoretical model (Figure 44A). From the
results, we can see that the propagation velocity and attenuation of Love waves are not sensitive
to storage shear modulus of cell layer with the range of 30-100KPa. With the increase of cell
viscosity, changes in propagation velocity and attenuation increase (Figure 44B). The sensitivity
of propagation velocity and attenuation to cell viscosity is higher than to the storage shear
modulus of cell layer. Figure 44C shows the propagation velocity and attenuation response to
the changes in the viscosity of the interfacial layer. We found that the device signals were much
124
sensitive to the properties of interfacial layer compared with the properties of cell layer. This is
reasonable since the interfacial layer is more close to the substrate; therefore, the acoustic wave
is stronger in the interfacial layer than in the cell layer.
5.6 SUMMARY
In summary, in this study, based on the steady state of cell adhesion to the solid surface, we
developed a multiple layer model to describe the structures of the cell-based Love mode acoustic
wave devices. According to this multiple layer model, the transfer matrix method was used to
determine the response of the device, including phase velocity and attenuation, to the changes in
physical parameters in the wave-guiding layer, interfacial layer and cell layer. According to
simulation results, it has been found that the sensitivity of the Love mode acoustic wave sensors
are affected by the phase velocity of piezoelectric substrate, and the phase velocity, viscosity and
thickness of wave-guiding layer. Therefore, our model can be used to optimize the device
structure and direct the selection of materials for the device to obtain best performance.
Experimentally, our results on the resonant frequency and insertion loss change with the
thickness of PMMA wave guiding layer showed that our theoretical models were valid to
describe the resonant frequency shift with loadings and insertion loss change induced by the loss
of loading materials in the Love mode acoustic wave device. Furthermore, we showed that
propagation velocity and attenuation shifts are sensitive to the properties and thickness of the
interfacial layer and cell layer, especially interfacial layer, which means the Love mode acoustic
wave devices provide potential tools to detect protein production, monitor cell adhesion. Since
our theoretical model cannot describe the signal attenuation induced by energy loss in bulk
125
material, in the future, we will improve our theoretical models by considering this part of loss in
them. In addition, we will perform the experimental studies on the use of Love mode acoustic
wave devices to detect cell behaviors based on the modeling results discussed in this paper.
126
6.0 THE APPLICATION OF SOFT LITHOGRAPHY TECHNIQUES TO
DETERMINE THE EFFECTS OF CELL SHAPE ON CELL STRUCTURES AND
FUNCTIONS
6.1 INTRODUCTION
In chapter 3, 4 and 5, I discussed the application of thickness shear mode (TSM) acoustic wave
sensors and Love mode acoustic wave sensors for cell biological studies. The other research
areas that are critical to the development of cell-based biosensors are the controlling of cell
function and fundamental studies of cell response to various hybrid interfaces.
In culture, cells attach to and spread on a substrate. Previous studies have shown that cell
attachment and spreading is related to gene expression and protein production, implying a cell
shape-based regulation of cellular function [39, 43, 112]. Understanding how cell shape
influences cellular function is important to regulating cell phenotype and developing engineered
tissues for medical applications [57, 113]. Therefore, in this chapter, I will discuss our research
on the controlling of cell shape by soft lithography and determining the effects of cell shape on
cell structure and function.
Soft lithography, a technique derived from microfabrication technology in the
semiconductor industry, provides a powerful approach for studying cell shape or spreading in
regulation of cell function [33, 39, 41, 114]. This technique allows fabrication of adhesive
127
patterns on the substrate at micro- to sub-micrometer scales. Confining cells to these adhesive
patterns with various shapes and sizes can precisely control cell shape and spreading area.
Micropatterning substrates by soft lithography overcomes the limitations of homogeneous
unconfined adhesive substrates on which cells display various shapes even though they belong to
the same cell type, which may partially alter their normal functions [115]. Using soft lithography,
previous studies show that the degree of cell spreading influences DNA synthesis [39],
proliferation and apoptosis [40, 41], and focal adhesion assembly [42] in human and bovine
capillary endothelial cells; differentiation in mesenchymal stem cells [43]; and gene expression
and protein synthesis in primary rat bone cells [44]. However, for a fixed spreading area, studies
on cell shape regulation are mainly limited to cell motility, i.e., lamellipodia extension and
direction of cell migration. For example, cells preferentially extended lamellipodia from their
corners regardless of their geometric shapes [45, 46]. During migration, narrow teardrop-shaped
or triangular cells moved predominantly toward the blunt end, while wide cells did not;
rectangular cells migrated outward towards their two ends without bias for either one, but
circular or square cells moved in random directions [33].
In this part of study, we aimed to determine the effect of cell shape on collagen
expression in human tendon fibroblasts (HTFs). In vivo, tendon fibroblasts are highly elongated
and are aligned along the longitudinal direction of the intact tendon, a well-organized fibrous
connective tissue that is mainly composed of parallel collagen fibers [116]. By producing
extracellular matrix (ECM) proteins, especially collagen type I, tendon fibroblasts serve to
maintain, repair, and remodel the tendon matrix [55]. As such, synthesis of collagen type I is one
major function of tendon fibroblasts. Therefore, the first aim of this study was to determine how
this particular elongated cell shape is related to collagen type I expression in HTFs.
128
It is known that cells generate mechanical tension in their actin cytoskeleton and exert
forces on cell-cell or cell-ECM contacts [77]. Tension in the actin cytoskeleton transmits to ECM
via focal adhesions, and this force is termed as cell traction. The alterations in cell traction forces
(CTFs) can result in change of cell shape [50, 76, 117]. Moreover, the physical connections
between focal adhesions, actin cytoskeleton, and nuclear scaffold is thought to provide a
pathway for mechanical signal transfer from cell surface to nucleus [50]. Therefore, the second
aim of this study was to test the hypothesis that change of collagen type I expression in HTFs
with various shapes is associated with alterations in the actin cytoskeletal structure and tension
distribution. Specifically, we examined the distribution of actin cytoskeleton, focal adhesions,
and CTFs in HTFs of various shapes and correlated them to the expression of collagen type I in
these tendon cells.
6.2 MATERIALS AND METHODS
6.2.1 Fabrication of micropatterned substrates
Micropatterned substrates were fabricated by microcontact printing method according to the
following procedures (Figure 45). First, silicon masters were fabricated by using standard
photolithography technologies. Negative photoresist SU-8 was spun on cleaned silicon wafer
(Figure 45a). The photoresist-coated silicon wafer was then exposed under UV light through the
patterned photomask on Suss MA 6 mask aligner for 20 sec (SUSS MicroTec Inc., Santa Clara,
CA) (Figure 45b). The features were then developed, and the silicon master with the desired
pattern was generated (Figure 45c). In order to facilitate subsequent release of
129
Poly(dimethylsiloxane) (PDMS) from the silicon master, the template was oxidized in air plasma
for 1 min and silanized with (tridecafluoro-1,1,2,2-tetrahydrooctyl)-1-trichlorosilane vapor
overnight under vacuum. PDMS was then prepared by thoroughly mixing two silicone
components, 601A and 601B (Wacker Chemical, Adrian, MI), in a weight ratio of 10:1, then
slowly poured onto the silicon master (Figure 45d). After curing at room temperature for 24 hr,
the PDMS was removed and cut to the desired size (Figure 45e). The substrate was prepared by
the following procedures. The glass slides were cleaned and coated with 10 nm thick chromium
and 20 nm thick gold (Figure 45f). The PDMS stamps were cleaned with ethanol and dried with
a stream of pressurized air. A 2 mM ethanol solution of tris(ethylene glycol)-terminated
alkanethiol (HS(CH2)11(OCH2CH2)3OH, ProChimia, Poland) was put onto the stamp surface and
incubated for 10 min, and the solution was then removed and the stamp was dried. The PDMS
stamp was then gently pressed onto the slide (Figure 45g), and after 1 min, the stamp was
removed from the glass substrate. A self-assembled monolayer was formed at the region where
tris(ethylene glycol)-terminated alkanethiol was stamped (Figure 45h). The substrate was then
rinsed with ethanol and dried with a steam of air. The self-assembled monolayer of tris(ethylene
glycol)-terminated alkanethiol can resist the protein adsorption. Therefore, cells only adhere to
the non-tris(ethylene glycol)-terminated alkanethiol-stamped region.
130
Si
SU-8
PDMS
Cr
Glass
Au
tris(ethylene glycol)-terminated alkanethiol
a
b
c
d
f
g
h
e
Si
SU-8
PDMS
Cr
Glass
Au
tris(ethylene glycol)-terminated alkanethiol
a
b
c
d
f
g
h
e
Figure 45. A schematic of microcontact printing method used for micropatterning gold surfaces in this study.
a
b
UV
PAG
Glass
PDMS
Type I collagen
Sulfo-SANPAH
ca
b
UV
PAG
Glass
PDMS
Type I collagen
Sulfo-SANPAH
c
Figure 46. A schematic of microcontact printing method used for micropatterning PA gel in this study.
131
Microcontact printing method was also been used for fabrication of adhesive islands on
polyacrylamide (PA) gel according to a method modified from the literature [118]. Briefly, after
the PA gel was fully cross-linked, Sulfo-SANPAH (Pierce, Rockford, IL), a heterobifunctional
cross-linker, was applied onto the gel surface and linked to the gel by exposing it to UV light
(Figure 46a). The PDMS stamps were treated with oxygen plasma and then dipped in 100 μg/ml
collagen type I solution (Angiotech Biomaterials, Palo Alto, CA). After carefully scraping the
solution off the top surface of stamp, the stamp with the solution of collagen filled in the holes
was gently pressed onto the PA gel overnight to allow conjugation of collagen type I to PA gel
surface (Figure 46b). The stamp was removed from gel surface and the gel was then rinsed with
PBS prior to cell seeding (Figure 46c).
6.2.2 Cell culture experiments
Human tendon fibroblast culture was established according to the published protocol [119]. For
immunofluorescence assays, human tendon fibroblasts (HTFs) were seeded onto patterned gold-
coated glass slides in cell growth medium (DMEM, 10% FBS and 1% penicillin/streptomycin).
After HTFs (passage 5 or 6) were cultured on the patterned surface for 36 hrs, cells were fixed
with 4% paraformaldehyde and then proceeded for immunostaining. For measuring CTFs, after
collagen type I was patterned onto the PAG, 3000 HTFs were plated onto each gel disk in
DMEM with 1% FBS and 1% penicillin/ streptomycin. Non-adhered cells were removed 1 hr
later, and fresh cell growth medium was added. Cells were then allowed to spread on PAG for 6
hr prior to image acquisition for CTF measurement. During cell culture, cells were maintained at
37 oC in humidified atmosphere with 95% air and 5% carbon dioxide.
132
6.2.3 Immunofluorescence assays for collagen type I, GAPDH, actin filaments, and vinculin
Collagen type I, actin filaments, vinculin, and glyceraldehyde 3-phosphate dehydrogenase
(GAPDH) were stained using the following protocols. Cells were fixed in 4% paraformaldehyde
(FD NeuroTechnologies, Ellicott City, MD) for 15 min and then blocked with a Blotto buffer
(4% dry milk and 0.4% Triton X-100 in PBS) for 45 min. For staining collagen type I and
GAPDH, the sample was incubated with goat polyclonal anti-type I collagen antibody (Santa
Cruz Biotechnology, Santa Cruz, CA) overnight or monoclonal antibody for GAPDH
(Biogenesis, UK) for 1h with a dilution ratio 1:100 and 1:300, respectively, and then washed
with block solutions. FITC-conjugated donkey anti-goat antibodies (Santa Cruz Biotechnology,
Santa Cruz, CA) were then applied at a dilution ratio of 1:250 for 1 hr. The sample was washed
extensively with block solutions and PBS to remove excess antibody. A FAK100 kit (Chemicon
International, Temecula, CA) was used to stain actin filaments and vinculin. Samples were
incubated with anti-vinculin with a dilution ratio 1:100 for 1 hr after cells were fixed and blocked
with a Blotto buffer. After samples were washed with block solutions, secondary antibody (goat
anti-mouse, FITC-conjugated, dilution 1:200) and TRITC-conjugated phalloidin (dilution 1:200)
were applied simultaneously for 1 hr at room temperature. Cell nuclei were then stained with
DAPI. Afterward, the sample was mounted and examined using fluorescence microscope or laser
confocal microscope.
6.2.4 Cell traction force microscopy (CTFM)
The determination of CTFs by CTFM was described previously [116]. It is based on the
deformations in the surface of a thin elastic substrate induced by cells cultured on it. The
133
fluorescent beads embedded in a thin PAG serve as markers for determining substrate
displacements by image analysis, which are then used to compute CTFs based on continuum
mechanics [76].
The PAG embedded with 0.5 μm red fluorescent micro-beads (Molecular Probes, Eugene,
OR) was prepared first. The gel disk was 120 μm in thickness, 10 mm in diameter, and attached
to the bottom of a 35 mm dish (MatTek, Ashland, MA), which had a circular glass area with a
diameter of 14 mm. The gel stiffness was controlled by changing the relative concentration of
acrylamide (Sigma, St. Louis, MO) and N, N'-methylenebisacrylamide (bis-acrylamide, Sigma,
St. Louis, MO). In this study, gel containing 5% acrylamide and 0.1% bis-acrylamide with a
Young’s modulus of 3 kPa and a Poisson’s ratio of 0.48 was used [57]. After collagen type I was
patterned onto the PAG using microcontact printing method and HTFs were cultured on those
gels for 6 h, phase contrast images of individual cells that were fully spread on single collagen-
coated islands and fluorescence images of the embedded fluorescent beads were taken. After
cells on the gel disk were trypsinized, fluorescence images of the fluorescent beads in the same
view were taken. The CTFs were then computed using a MATLAB program based on a
published method [116].
6.2.5 Quantification of collagen type I expression levels
The total collagen type I expression level per cell was quantified as follows. The
immunofluorescence microscopy images of cells with different shapes were taken under the
same conditions, including the same exposure time and gain factor. Cells that were fully spread
on entire islands, as confirmed by phase contrast images, were selected for analysis. Using the
image analysis software, (MetaMorph software. Molecular Devices, Sunnyvale, CA), the
134
boundary of the chosen cell was defined manually first. The intensities per pixel in fluorescence
microscopy image were automatically measured and the average intensity over the defined cell
area was computed. To subtract the background noise, we defined four small areas near cell
boundary. By subtracting the average intensity over these the four small areas, we obtained the
immunofluorescence intensity level for each chosen cell. A total of 20 cells from each group
were chosen for measurement. To verify that cell shape affected the collagen type I synthesis, we
also performed immunocytochemistry for GAPDH and measured its fluorescence intensity of
each cell for four different shapes.
6.2.6 Statistical analysis
A one-way analysis of variance (ANOVA) was used to analyze the effect of cell shape on
collagen type I expression level and CTF magnitude, followed by Duncan’s test for multiple
comparisons with the significance level set at α = 0.05.
6.3 RESULTS
6.3.1 Micropatterning human tendon fibroblasts with different shapes
Micropatterned substrates containing an array of cell-adhesive islands surrounded by PEG-
covered non-adhesive regions were fabricated using microcontact printing technique. The
adhesive islands were designed to be of the same area (1.96×103 µm2) but in different geometric
forms (rectangular with aspect ratios of 19.6, 4.9 and 2.2, respectively, and circular). This area is
135
comparable with typical spreading area of human tendon fibroblasts (HTFs) on a smooth surface.
To prevent cells from spreading across multiple islands, the non-adhesive spacing between
islands was designed to be 50 µm. When HTFs were seeded onto the patterned substrates, they
preferentially attached to adhesive islands and were confined within the boundaries of these
islands. Therefore, the shape of cells was as precisely controlled as that of adhesive islands (Fig.
47A). In order to measure traction forces of micropatterned cells, type I collagen-coated adhesive
islands on PAGs were fabricated. The shape of cells was seen to be well controlled on PAG
surface (Fig. 47B).
AA BB
Figure 47. HTFs on (A) Micropatterned gold-coated substrates and (B) micropatterned polyacrylamide gel. (Scale bars: 50 μm).
6.3.2 Expression levels of collagen type I in micropatterned HTFs
Expression of collagen type I in HTFs with different shapes was determined by
immunofluorescence assay. The fluorescence microscopy images of collagen type I for cells with
different shapes (Fig. 48) showed that fluorescence intensity decreased with decreasing the
aspect ratio of cells. This observation was confirmed by quantification by using image analysis
136
(Fig. 49). Compared to rectangular cells with an aspect ratio of 19.6 (i.e. the most elongated
cells), cells with an aspect ratio of 4.9 expressed about 10% lower collagen type I, while cells
with an aspect ratio of 2.2 expressed more than 20% lower. For circular cells, whose aspect ratio
is 1, the collagen expression was 40% lower than that of the most elongated cells with an aspect
ratio of 19.6. However, we found that GAPDH expression levels were not significantly different
among cells with the four shapes (data not shown).
H
A B
C
D
Figure 48. Immunofluorescence microscopy images of collagen type I expression in rectangular HTFs with aspect ratios of 19.6 (A), 4.9 (B), and 2.2 (C), respectively, and circular HTF (D). (Scale bars: 20 μm).
6.3.3 Organization of actin filaments and focal adhesions in micropatterned HTFs
Long and parallel actin filament bundles were formed in elongated cells, with aspect ratios of
19.6 and 4.9, respectively, along the longitudinal direction, while in short cells with an aspect
ratio of 2.2 and circular cells, they were along the circumference of cells, and several bundles
137
were formed from the nuclear area to the edge of cells (Figure 49). Focal adhesions, revealed by
staining vinculin, were concentrated at the two ends of elongated cells or corners of short cells.
In circular cells, however, focal adhesions were uniformly distributed along the edge of cell.
Merged images of F-actin and vinculin showed that F-actin filaments (or stress fibers) at their
ends linked with focal adhesions, where cells anchor to underlying substrate and hence exert
traction forces.
19.6 4.9 2.2 1Aspect ratio
Nor
mal
ized
Fl
uore
scen
ce In
tens
ity
0
0.4
0.8
1.2
1.6*
* p < 0.05 (n = 20)
*
19.6 4.9 2.2 1Aspect ratio
Nor
mal
ized
Fl
uore
scen
ce In
tens
ity
0
0.4
0.8
1.2
1.6*
* p < 0.05 (n = 20)
*
Figure 49. Effect of cell shape on collagen type I expression level.
6.3.4 Cell traction forces in micropatterned HTFs
For all cells examined in this study, the largest CTFs were located at the two ends of elongated
cells, at the corner regions of short cells, and along the outer edge of the circular cells. The
traction force directions appeared to be parallel to stress fibers. For example, the directions of the
largest CTFs were in parallel to the long axis in elongated cells, along the diagonal in short cells,
and along the radial in circular cells (Fig. 50). The magnitude of CTFs decreased with increasing
the aspect ratio of cell shape (Fig. 51).
138
250
200
150
100
50
400300200100
400
300
200
100
F-actin Vinculin Merge CTF200
160
120
80
40
A
B
C
D
250
200
150
100
50
400300200100
400
300
200
100
F-actin Vinculin Merge CTF200
160
120
80
40
A
B
C
D
Figure 50. Immunofluorescence microscopy images of F-actin, vinculin, and the overlay of them, as well as CTFs in micropatterned HTFs. A-C. Rectangular HTFs with aspect ratios of 19.6, 4.9, and 2.2, respectively; D circular
HTF. (Scale bars: 20 μm).
139
0
50
100
150
200
250
19.6 4.9 2.2 1Aspect ratio
Cel
l Tra
ctio
n Fo
rce
(Pa)
300
*
* p < 0.05 (n = 10)
**
Figure 51. The effect of cell shape on CTF magnitude. It increased with decreasing aspect ratio of the cell.
6.4 DISCUSSION
Micropatterned substrates allow us to control the shape and projection area of cells precisely. As
shown in this study, human tendon cells can be confined on micropatterned adhesive islands of
any shape and size as desired. Therefore, micropatterned cell-adhesive surface provides an ideal
experimental platform for studying functions of cells that have specific shapes in vivo at whole
cell and sub-cellular levels [34]. Moreover, precisely controlling cell shape reduces cell
variability as opposed to a cell population such that the biological response of the single cell can
be investigated, instead of “averaged response” of a heterogeneous cell population. For example,
micropatterned substrates have been used to create cell-based sensors in which cells exhibit a
140
more reproducible and defined phenotype [67, 120]. It should be noted that, however, while
micropatterning technology can control cell shape precisely and therefore creates a more uniform
cell group compared to cells on non-micropatterned surfaces, it cannot control cell cycle; in other
words, these micropatterned cells may be in different cell cycles; as a result, these
micropatterned cells can still differ in their functions.
In vivo, tendon cells, which are aligned with the collagen fibers in the longitudinal
direction of the tendons, are highly elongated in shape [116]. By producing collagen type I and
other matrix proteins, tendon cells in vivo function to maintain, repair, and remodel tendon
matrix [119]. Using micropatterned HTFs, this study was able to show collagen type I expression
levels at single cell level. Furthermore, collagen type I expression was the highest in the most
elongated cells, which have shape most similar to those tendon cells in vivo, suggesting that
normal tendon fibroblasts in vivo have an “optimal shape” for maximizing synthesis of type I
collagen, a major function of these cells. For those HTFs assuming circular shape, their collagen
type I expression levels were, on average, only ~65% compared to that of those elongated cells.
The reduction in collagen expression in tendon cells of circular shape may contribute to the
development of tendinopathy (Wang et al. 2006). Based on these data, it is also tempting to
suggest that these circular cells may behave like fibrocartilage cells, as they are round in vivo and
are known to produce low level of collagen type I but high level of collagen type II instead [34,
121].
An immediate question is: How does cell shape link to the biochemical process of protein
synthesis in a cell? It is known that mechanical stresses generated by cells within cytoskeleton
play important roles in the control of cell shape as well as cell function [50, 117]. Many
signaling molecules bind to the cytoskeleton and/or link with integrins at focal adhesion
141
complexes, both of which are highly mechanical deformation-sensitive [43, 122-124]. Therefore,
deformation in cytoskeleton, regulated by the mechanical stresses in it, provides a physical basis
for transducing mechanical signals into biochemical signals [125]. In addition, because of the
physical linkage between cytoskeleton (e.g. intermediate filaments) and nuclear matrix,
deformation of nuclear matrix caused by cytoskeleton reorganization will alter DNA topography
and transcription and consequently affect cell functions, such as gene expression and protein
synthesis [123].
Besides the cytoskeleton, cell-generated cytoskeletal tension influences cell fate and
functions [43, 51]. The changes in cell-generated forces have been shown to be responsible for
the effect of cell spreading area on cell proliferation [39, 41, 126]. Cells with larger spreading
area generate stronger tension and have an enhanced ability to proliferate [39, 41, 46]. Tension in
the cytoskeleton has also been shown to be critical in the commitment of human mesenchymal
stem cells into adipocyte or osteoblast lineages [43]. Using cells populated in collagen gels,
Agarwal et al. showed that higher contraction was related to higher levels of collagen production
in fibroblasts [127]. In this study, we found that changes in cell shape resulted in different
organizations in cytoskeletal structures and altered the direction and spatial CTF distribution.
This may be responsible for our finding that tendon cells with various aspect ratios expressed
different levels of collagen type I. In addition, given the fact that cells can use traction forces to
reorganize collagen fibers [128, 129] and elongated cells produced traction forces along their
long axis, we speculate that this may be one reason why these elongated tendon cells produce
aligned collagen matrix in vivo and in vitro [116].
In this study, we have also demonstrated that cell shape, controlled by micropatterned
adhesive islands on substrate, influenced spatial distribution of actin cytoskeleton, focal
142
adhesions, and direction and spatial CTF distribution. The mechanisms by which cell shape
alters actin filament organization and spatial arrangement of focal adhesions and traction forces
might be due to the regulation of mechanical signals from the ECM-coated islands with various
shapes during cell spreading. As shown in this study, the largest traction stresses occurred at the
two ends of highly elongated rectangular cells, at corners of shortened rectangular cells, or
around the edge of circular cells. These mechanical stresses may promote integrin clustering and
focal adhesion complex formation, resulting in cytoskeletal realignment in the formation of
bundles of contractile stress fibers along the directions of stresses stretching from focal
adhesions, as well as recruitment of myosin to generate cell contraction [115, 117]. Finally, cells
establish a stabilized structure of cytoskeleton and focal adhesions against ECM and achieve
force balance between cells and ECM. Physical connections between integrins, cytoskeletal
filaments, and nuclear scaffolds, which have been observed in our immunostaining results and
also demonstrated in previous studies [50, 117], may provide a pathway for the mechanical
signal transfer throughout cells, thereby producing integrated changes in cellular structure in
response to changes in cell-ECM adhesion [123]. However, the exact molecular mechanisms that
are responsible for cell shape-regulated cellular function seen in this study are not known. It
could be that alterations in CTFs and other physical effects resulting from the imposed shape
serve as the actual signals that trigger a cascade of molecular events leading to changes of
cellular function. Further studies are required to control cell cycle and probe those molecular
signals that account for alteration of cellular function with different shapes.
In summary, using micropatterning techniques, we have investigated the effect of cell
shape on the level of collagen type I expression as well as on the changes of actin cytoskeleton,
focal adhesions, and CTFs. We have shown that elongated tendon cells expressed higher
143
collagen type I than those shorter cells even though cell spreading area was constant, indicating
tendon cells in vivo, which assume an elongated shape in normal tendons, has an “optimal shape”
to perform their major function, that is, to synthesize collagen type I. We have also shown that
changes in cell shape induced alterations in actin cytoskeletal structure, spatial arrangement of
focal adhesions, direction and spatial distribution of traction forces in tendon cells, which may
collectively affect the observed differential collagen type I expression in tendon cells with
different shapes. Future studies are necessary to investigate the molecular signaling events that
are responsible for cell shape-regulated cellular function.
144
7.0 CONCLUDING REMAKS
The research in this dissertation is focused on investigating the cell biological applications of
acoustic wave sensors and the effects of cell shape on cell structure and function. Both of these
two parts of research are very important to the realization of cell-based biosensors in various
applications.
The thickness shear mode acoustic wave cytosensor system is developed to monitor the
procedure of cell attachment, spreading and formation of focal adhesion to solid surface after
cells added into the cell medium and cell detachment under cytoskeleton disruption, and to
characterize the viscoelastic properties of cell monolayer after cell adhesion to sensor surface
reaches the steady state. This part of work includes theoretical modeling, design of the sensor
system and the experimental studies for biological application of this sensor system. Our studies
expanded the cell biological applications of thickness shear mode acoustic wave sensors.
It has been recognized that acoustic wave devices with a higher operation frequency
have a higher sensitivity. Surface acoustic wave devices are able to obtain high frequency more
easily than bulk acoustic wave devices. Therefore, in this dissertation, besides the bulk acoustic
wave sensors, we studied the surface acoustic wave sensors. Our studies are focused on the
theoretical modeling of Love mode surface acoustic wave biosensors. Basically, we developed a
theoretical model to describe how electrical signals of the Love mode acoustic wave sensors
change with various multilayer loadings. Our experimental results demonstrated the validity of
145
our model. Through our theoretical models, we discussed how to optimize the design of Love
mode acoustic wave sensors in order to obtain high sensitivity and predicted the electrical signal
change when the surface of the device is attached with a cell layer. This part of research is
critical to the future experimental studies of Love mode acoustic wave devices as cell-based
biosensors.
In terms of the studies on controlling cell function, we precisely controlled cell shape
using soft lithography, which is a technique derived from the microfabrication technology. By
micro-patterning cells, we studied the effects of cell shape on cytoskeleton structure, focal
adhesion morphology, cell traction force and protein synthesis by cells. This part of research is
critical to improving the sensitivity and reliability of cell-based biosensors and developing
engineered tissues for biomedical applications.
7.1 SUMMATY OF ACCOMPLISHMENTS
The primary accomplishments of this dissertation are listed below.
Theoretical models of uniform cell monolayer and non-uniform cell adhesion to the
thickness shear mode resonator were developed. These models first time predict the
relationship between the electrical signals of the device (resonate frequency, motional
resistance and electrical impedance or admittance) and mechanical properties and thickness
of the interfacial layer and the cell layer. They are critical to explaining the experimental
data, which allows us to obtain more information on cell-substrate contact and cell
viscoelasticity. In addition, using these theoretical models, we first time characterized the
viscoelasticity of cell monolayer by thickness shear mode acoustic wave sensors.
146
The procedure of cell attachment, spreading and formation of focal adhesion with different
cell seeding densities to solid surface is monitored by using the thickness shear mode
acoustic wave sensor system. Our results showed that thickness shear mode acoustic wave
devices could be used to monitor the procedure of cell attachment and spreading. The
alterations in viscoelastic properties and cell-substrate contact during the formation of focal
adhesion can be detected by the thickness shear mode sensors also.
Changes in cell adhesion with cytoskeleton disruption are investigated using the thickness
shear mode acoustic wave cytosensor system. Our results showed that changes in motional
resistance are dependent on the concentration of cytochalasin D, which demonstrated that
the thickness shear mode sensors can be used to quantify the effect of drug on cell behaviors.
The viscoelastic properties of cell monolayer are determined based on the theoretical model
of the uniform cell adhesion and curve-fitting algorithmic strategy. Our research first time
demonstrated that thickness shear mode acoustic wave sensor provided a reliable tool to
characterized cell viscoelasticity in a non-invasive, fast and simple way.
A theoretical model of the Love mode surface acoustic wave sensor to predict the changes in
phase velocity and loss in the acoustic waves with the thickness and viscoelasticity of
loading layers is developed. Our experimental studies demonstrated the validity of our
model. Based on the theoretical model, the wave-guiding layer can be optimized to obtain
the high sensitivity of the device. Changes in phase velocity and loss of the acoustic waves
in the wave-guided shear-horizontal surface acoustic wave sensor with cell monolayer
adhesion are predicted based on the theoretical model. This part of research will be critial to
the future experimental studies on Love mode acoustic wave sensors for cell biological
applications.
147
Adhesive proteins are micro-patterned on various substrates by soft lithography techniques.
The effect of cell shape on cytoskeleton structure, focal adhesion morphology, cell traction
force and protein synthesis is investigated experimentally. It has been found that the
elongated tendon cells express higher collagen type I than those shorter cells even though
cell spreading are is constant. It has also shown that changes in cell shape induced
alterations in cytoskeleton structure, spatial arrangement of focal adhesions, traction forces
in tendon cells, which may collectively affect the observed differential collagen type I
expression in tendon cells with different shapes. The techniques of controlling cell shape on
various substrates can be used in the development of cell-based biosensors in the future. The
regulation mechanism of cell shape on cell function provides design criteria for the
development of cell-based biosensors to preserve the biological response of interests.
7.2 FUTURE PERSPECTIVE
The work on the TSM devices for characterizing cell mechanical properties could be continued
in the future. Based on our current model for cells adhered to the TSM resonator, there are too
many unknown parameters in this model. Therefore, I propose to combine the TSM resonator
with electric cell-substrate impedance sensing or surface plasmon resonance technology to
measure the cell-substrate interaction, which will help us obtain enough information for
extracting cell mechanical properties. This sensor system could be used for characterizing
mechanical properties for different cell lines, normal or diseased cells and cells under various
biological environments, which have potential applications in biomedical research, drug
screening etc.
148
According to our current results, the Love mode acoustic wave devices could be used for
measuring the interfacial layer between cells and substrate. It is expected that this device can be
used to monitor the assembling of cell focal adhesion and extracellualr matrix production in real
time, which is critical in the physiological and pathological processes.
Based on our previous studies, we demonstrated that understanding the relationship
between cell shape and cellular function is important for study of cell biology in general and for
regulation of cell phenotype in tissue engineering in particular. In the future work, cell shape will
be controlled using soft lithography on the sensor surface; therefore, the effect of cell shape on
cell behaviors can be studied using our sensor systems. The results will be compared with those
by conventional biological techniques, such as immunoassaying, western blot etc. Compared
with those conventional biological techniques, the sensor system has advantages of fast, simple,
real-time and quantitative in measurements.
149
BIBLIOGRAPHY
1. Schultz, J., et al., WTEC Panel Report on INTERNATIONAL RESEARCH AND DEVELOPMENT IN BIOSENSING. 2004.
2. Ballantine, D.S. and R.M. White, Acoustic Wave Sensors: Theory, Design, and Physico-Chemical Applications.
3. Li, Z., Guided shear-horizontal surface acoustic wave (SH-SAW) chemical sensors for detection of organic contaminants in aqueous environments. Ph.D thesis. 2005.
4. Bunde, R.L. and E.J. Jarvi, Piezoelectric quartz crystal biosensors. Talanta., 1998. 46: p. 1223-1236.
5. Rogers, K.R.a.C.L.G., Environmental Biosensors: A Status Report. http://pubs.acs.org/hotartcl/est/96/nov/envir.html.
6. Beck, K., et al., Contactless surface acoustic wave gas sensor Sensors and actuators A: Physical, 1999. 76(1-3): p. 103-106.
7. Cheeke, J., N. Tashtoush, and N. Eddy, Surface Acoustic Wave Humidity Sensor Based on the Changes in the Viscoelastic Properties of a Polymer Film. Proc Ultrasonics Symposium, , 1996. 1: p. 449-452.
8. Thompson, M., Stone, D. C., Surface-launched acoustic wave sensors: chemical sensing and thin film characterization.
9. Larson, P.H., The Use of Piezoelectric Material in Creating Adaptive Shell Structures.” Ph. D. thesis. 1994.
10. Herrmann, F. and B. Jakoby, Microacoustic sensors for liquid monitoring. sensor update 2001. 9: p. 105-160.
11. Vellekoop, M.J. and A.J. Van Rhijn, All-silicon plate wave oscillator system for sensors. Sensor and actuators A, 1991. 27: p. 699-703.
12. Lucklum, R. and P. Hauptmann, Transduction mechanism of acoustic-wave based chemical and biochemical sensors. Meas. Sci. Technol, 2003. 14: p. 1854-1864.
150
13. Berzina, T.S., L. Piras, and V.I. Troitsky, Study of horseradish peroxidase activity in alternate-layer Langmuir–Blodgett films. Thin Solid Films, 1998. 327–329 p. 621-626.
14. Rickert, J., A. Brecht, and W. Gopel, Quartz crystal microbalances for quantitative biosensing and characterizing protein multilayers. Biosens. Bioelectron. , 1997. 12: p. 567–575.
15. Lvov, Y., et al., Formation of ultrathin multilayer and hydrated gel from montmorillonite and linear polycations. Langmuir, 1996. 12: p. 3038–3044.
16. Ko¨nig, B. and M. Gra¨tzel, A quartz crystal microbalance assay for detection of antibodies against the recombinant African swine fever virus attachment protein p12 in swine serum. Anal. Chem., 1994. 66: p. 341–344.
17. Koenig, B. and M. Graetzel, A novel immunosensor for herpes viruses. Anal. Chem. , 1994. 66: p. 341-344.
18. Nakanishi, K., H. Muguruma, and I. Karube, A novel method of immobilizing antibodies on a quartz crystal microbalance using plasma-polymerized films for immunosensors. Anal. Chem. , 1998. 68: p. 1695–1700.
19. Carter, R.M., et al., Quartz crystal microbalance detection of Vibrio cholerae O139 serotype. J. Immunol. Methods 1995. 187: p. 121–125.
20. Fredriksson, C., et al., The piezoelectric quartz crystal mass and dissipation sensor: a means of studying cell adhesion. Langmuir 1998. 14(2): p. 248 -251. .
21. Zhou, T., et al., The quartz crystal microbalance as a continuous monitoring tool for the study of endothelial cell surface attachment and growth. Biotechnology Progress, 2000. 16(2): p. 268-277.
22. Marx, K.A., et al., Quartz crystal microbalance study of endothelial cell number dependent differences in initial adhesion and steady-state behavior: evidence for cell-cell cooperativity in initial adhesion and spreading. Biotechnology Progress, 2003. 19: p. 987-999.
23. Wegener, J., et al., Analysis of the composite response of shear wave resonators to the attachment of mammalian cells. Biophysical Journal, 2000 78: p. 2821–2833.
24. Redepenning, J., et al., Osteoblast attachment monitored with a quartz crystal microbalance. Anal Chem, 1993. 65(23): p. 3378-81.
25. Wegener, J. and A. Janshoff, Cell adhesion monitoring using a quartz crystal microbalance: comparative analysis of different mammalian cell lines. The European Biophysics Journal, 1998. 28: p. 26-37.
26. Kalantar-Zadeh, K., et al., A Novel Love mode surface acoustic wave based immunosensors. Sensors and Actuators B: Chemical, 2003. 91(143-147).
151
27. McHale, G., M.I. Newton, and F. Martin, Theoretical mass sensitivity of Love wave and layer guided acoustic plate mode sensors. J. Appl. Phys., 2002. 91(12): p. 9701-9710.
28. Li, Z., et al., Design Considerations for High Sensitivity Guided SH-SAW Chemical Sensor for Detection in Aqueous Environments. 2004 IEEE International Ultrasonics, Ferroelectrics, and Frequency Control Joint 50th Anniversary Conference, 2004: p. 185-192.
29. Martin', S.J. and G.C. Frye, Dynamics and Response of Polymer-Coated Surface Acoustic Wave Devices: Effect of Viscoelastic Properties and Film Resonance. Anal. Chem., 1994. 66: p. 2201-2219.
30. Oliner, A.A., Microwave network methods for guided elastic waves. IEEE Transactions on microwave theory and techniques, 1969. MTT-17(11): p. 812-826.
31. Francis, L.A., Thin film acoustic waveguides and resonators for gravimetric sensing applications in liquid. Ph. D. Thesis, Universit´e catholique de Louvain, Louvain-la-Neuve, Belgium, 2006.
32. Xia, Y. and G.M. Whitesides, Soft lithography. Annu. Rev. Mater. Sci., 1998. 28: p. 153-84.
33. Whitesides, G.M., et al., Soft lithography in biology and biochemistry. Annu Rev Biomed Eng, 2001. 3: p. 335-73.
34. Lehnert, D., et al., Cell behaviour on micropatterned substrata: limits of extracellular matrix geometry for spreading and adhesion. J Cell Sci, 2004. 117(Pt 1): p. 41-52.
35. Xia, Y., et al., Complex Optical Surfaces Formed by Replica Molding Against Elastomeric Masters. Science, 1996. 273(5273): p. 347-9.
36. Kim, E., Y. Xia, and G.M. Whitesides, Polymer microstructures formed by moulding in capillaries. Nature, 1995. 376: p. 581-584.
37. Joon, N.L., et al., Patterning of dielectric oxide thin layers by microcontact printing of self-assembled monolayers. J. Mat. Res., 1995. 10: p. 2996-99.
38. Wang, D., et al., Nanometer scale patterning and pattern transfer on amorphous Si, crystalline Si, and SiO2 surfaces using self-assembled monolayers. Appl. Phys. Lett., 1997. 70: p. 1593-95.
39. Singhvi, R., et al., Engineering cell shape and function. Science, 1994. 264(5159): p. 696-8.
40. Chen, C.S., et al., Micropatterned surfaces for control of cell shape, position, and function. Biotechnol Prog, 1998. 14(3): p. 356-63.
41. Chen, C.S., et al., Geometric control of cell life and death. Science, 1997. 276(5317): p. 1425-8.
152
42. Chen, C.S., et al., Cell shape provides global control of focal adhesion assembly. Biochem Biophys Res Commun, 2003. 307(2): p. 355-61.
43. McBeath, R., et al., Cell shape, cytoskeletal tension, and RhoA regulate stem cell lineage commitment. Dev Cell, 2004. 6(4): p. 483-95.
44. Thomas, C.H., et al., Engineering gene expression and protein synthesis by modulation of nuclear shape. Proc Natl Acad Sci U S A, 2002. 99(4): p. 1972-7.
45. Brock, A., et al., Geometric determinants of directional cell motility revealed using microcontact printing. Langmuir, 2003. 19(5): p. 1611-7.
46. Parker, K.K., et al., Directional control of lamellipodia extension by constraining cell shape and orienting cell tractional forces. Faseb J, 2002. 16(10): p. 1195-204.
47. Bray, D., Cell Movements. 1992.
48. Lodish, H. and D. Baltimore, Molecular Cell Biology. 1995.
49. Ingber, D.E., Tensegrity I. Cell structure and hierarchical systems biology. J Cell Sci, 2003. 116(Pt 7): p. 1157-73.
50. Maniotis, A.J., C.S. Chen, and D.E. Ingber, Demonstration of mechanical connections between integrins, cytoskeletal filaments, and nucleoplasm that stabilize nuclear structure. Proc Natl Acad Sci U S A, 1997. 94(3): p. 849-54.
51. Wang, N., et al., Mechanical behavior in living cells consistent with the tensegrity model. Proc Natl Acad Sci U S A, 2001. 98(14): p. 7765-70.
52. Ingber, D.E., The riddle of morphogenesis: a question of solution chemistry or molecular cell engineering? Cell, 1993. 75(7): p. 1249-52.
53. Ingber, D.E., Tensegrity II. How structural networks influence cellular information processing networks. J Cell Sci, 2003. 116(Pt 8): p. 1397-408.
54. Agutter, P.S., Cell mechanics and stress: from molecular details to the 'universal cell reaction' and hormesis. Bioessays, 2007. 29(4): p. 324-33.
55. Wang, J.H. and B.P. Thampatty, An introductory review of cell mechanobiology. Biomech Model Mechanobiol, 2006. 5(1): p. 1-16.
56. Khatiwala, C.B. and S.R. Peyton, The intrinsic mechanical properties of the extracellular matrix affect the behavior of preosteoblastic cells. Am. J. Physiol., in press.
57. Liu, W.F. and C.S. Chen, Cellular and multicellular form and function. Adv Drug Deliv Rev, 2007. 59(13): p. 1319-28.
58. Geiger, B. and A. Bershadsky, Transmembrane extracelluar matrix-cytoskeleton crosstalk. Nature review, 2001. 2: p. 793.
153
59. Sniadecki, N.J. and R.A. Desai, Nanotechnology for Cell–Substrate Interactions. Annals of Biomedical Engineering, 2006. 34(1): p. 59–74.
60. Murphy-Ullrich, J.E., The de-adhesive activity of matricellular proteins: is intermediate cell adhesion an adaptive state? J Clin Invest, 2001. 107(7): p. 785-90.
61. Lo, C.-M. and J. Ferrier, Impedance analysis of fibroblastic cell layers measured by electric cell-substrate impedance sensing. Physical Review E., 1998. 57(6): p. 6982-6987.
62. Lo, C.M., et al., Cell-substrate separation: effect of applied force and temperature. European Biophysics Journal 1998. 27(1): p. 9-17.
63. http://www.btxonline.com/products/ecis/.
64. Yamada, S., D. Wirtz, and S.C. Kuo, Mechanics of living cells measured by laser tracking microrheology. Biophysical Journal, 2000. 78(4): p. 1736-47.
65. Bausch, A.R., et al., Local measurements of viscoelastic parameters of adherent cell surfaces by magnetic bead microrheometry. Biophys J, 1998. 75(4): p. 2038-49.
66. Maksym, G.N., et al., Mechanical properties of cultured human airway smooth muscle cells from 0.05 to 0.4 Hz. Journal of Applied Physiology 2000. 89(4): p. 1619-32.
67. Park, S., et al., Cell motility and local viscoelasticity of fibroblasts. Biophys J, 2005. 89(6): p. 4330-42.
68. Trepat, X., et al., Viscoelasticity of human alveolar epithelial cells subjected to stretch. Am J Physiol Lung Cell Mol Physiol, 2004. 287(5): p. L1025-34.
69. Wu, Z.Z., et al., Comparison of the viscoelastic properties of normal hepatocytes and hepatocellular carcinoma cells under cytoskeletal perturbation. Biorheology, 2000. 37(4): p. 279-90.
70. Loh, O., A. Vaziri, and H.D. Espinosa, The Potential of MEMS for Advancing Experiments and Modeling in Cell Mechanics. Experimental Mechanics, 2007. DOI 10.1007/s11340-007-9099-8.
71. Rand, R.P., Mechanical Properties of the Red Cell Membrane. Ii. Viscoelastic Breakdown of the Membrane. Biophys J, 1964. 4: p. 303-16.
72. Suresh, S., et al., Connections between single-cell biomechanics and human disease states: gastrointestinal cancer and malaria. Acta Biomaterialia, 2005. 1: p. 15-30.
73. Thoumine, O. and A. Ott, Time scale dependent viscoelastic and contractile regimes in fibroblasts probed by microplate manipulation. J Cell Sci, 1997. 110 ( Pt 17): p. 2109-16.
74. Mahaffy, R.E., et al., Quantitative analysis of the viscoelastic properties of thin regions of fibroblasts using atomic force microscopy. Biophys J, 2004. 86(3): p. 1777-93.
154
75. Fabry, B., et al., Scaling the microrheology of living cells. Phys Rev Lett, 2001. 87(14): p. 148102.
76. Wang, J.H. and J.S. Lin, Cell traction force and measurement methods. Biomech Model Mechanobiol, 2007. 6(6): p. 361-71.
77. Harris, A.K., D. Stopak, and P. Wild, Fibroblast traction as a mechanism for collagen morphogenesis. Nature, 1981. 290(5803): p. 249–251.
78. Galbraith, C.G. and M.P. Sheetz, A micromachined device provides a new bend on fibroblast traction forces. Proc Natl Acad Sci USA, 1997. 94(17): p. 9114–9118.
79. du Roure, O., et al., Force mapping in epithelial cell migration. Proc Natl Acad Sci U S A, 2005. 102(7): p. 2390-5.
80. Tan, J.L., et al., Cells lying on a bed of microneedles: an approach to isolate mechanical force. Proc Natl Acad Sci U S A, 2003. 100(4): p. 1484-9.
81. Munevar, S., Y. Wang, and M. Dembo, Traction force microscopy of migrating normal and H-ras transformed 3T3 fibroblasts. Biophys J, 2001. 80(4): p. 1744-57.
82. Lee, J., The use of gelatin substrates for traction force microscopy in rapidly moving cells. Methods Cell Biol, 2007. 83: p. 297-312.
83. Lombardi, M.L., et al., Traction force microscopy in Dictyostelium reveals distinct roles for myosin II motor and actin-crosslinking activity in polarized cell movement. J Cell Sci, 2007. 120(Pt 9): p. 1624-34.
84. Kino, G.S., Acoustic Waves: Devices, Imaging and Analog Signal Processing, Prentice-Hall Englewood Cliffs, NJ. 1987.
85. Lucklum, R. and P. Hauptmann, Generalized acoustic parameters of non-homogenous thin films. 2002 IEEE international frequency control symposium and PDA exhibition, 2002: p. 234-241.
86. Rosenbaum, J.F., Bulk Acoustic Wave Theory and Devices Artech House Publishers, Boston, MA. 1988.
87. Chen, Q. and Q.-M. Wang, The effective electromechanical coupling coefficient of piezoelectric thin film resonators. Applied Physics Letters 2005. 86: p. 022904
88. Behling, C., R. Lucklum, and P. Hauptmann, The non-gravimetric quartz crystal resonator response and its application for polymer shear moduli determination. Measurement Science and Technology, 1998. 9: p. 1886-1893.
89. Lucklum, R., C. Behling, and P. Hauptmann, Role of mass accumulation and viscoelastic film properties for the response of acoustic-wave-based chemical sensors Analytical Chemistry, 1999. 71 p. 2488-2496.
155
90. Lucklum, R. and P. Hauptmann, Transduction mechanism of acoustic-wave based chemical and biochemical sensors Measurement Science and Technology 2003. 14: p. 1854-1864.
91. Becker, W.M. and D.W. Deamer, The World of the Cell, 2nd ed. 1991.
92. Giebel, K., et al., Imaging of cell/substrate contacts of living cells with surface plasmon resonance microscopy. Biophysical Journal, 1999. 76(1 Pt 1): p. 509-16.
93. Wu, H.W., T. Kuhn, and V.T. Moy, Mechanical properties of L929 cells measured by atomic force microscopy: effects of anticytoskeletal drugs and membrane crosslinking. Scanning, 1998. 20(5): p. 389-97.
94. Girard, K.D., et al., Dynacortin contributes to cortical viscoelasticity and helps define the shape changes of cytokinesis. The EMBO Journal, 2004. 23(7): p. 1536-46.
95. Sapper, A., J. Wegener, and A. Janshoff, Cell motility probed by noise analysis of thickness shear mode resonators. Analytical Chemistry, 2006. 78(14): p. 5184-91.
96. Smith, B.A., et al., Probing the viscoelastic behavior of cultured airway smooth muscle cells with atomic force microscopy: stiffening induced by contractile agonist. Biophys J, 2005. 88(4): p. 2994-3007.
97. Bausch, A.R., W. Moller, and E. Sackmann, Measurement of local viscoelasticity and forces in living cells by magnetic tweezer. Biophys. J. , 1999. 76: p. 573-579.
98. Hochmuth, R.M., Measuring the mechanical properties of individual human blood cells. J Biomech Eng, 1993. 115(4B): p. 515-9.
99. Felder, S. and E.L. Elson, Mechanics of fibroblast locomotion: quantitative analysis of forces and motions at the leading lamellas of fibroblasts. J. Cell. Biol. , 1990. 111: p. 2513-252.
100. Petersen, N.O., W.B. McConnaughey, and E.L. Elson, Dependence of locally measured cellular deformability on position on the cell, temperature, and cytochalasin B. Proc Natl Acad Sci U S A, 1982. 79(17): p. 5327-31.
101. Jimenez, Y., et al., A contribution to solve the problem of coating properties extraction in quartz crystal microbalance applications. IEEE Trans Ultrason Ferroelectr Freq Control, 2006. 53(5): p. 1057-72.
102. Feneberg, W., M. Aepfelbacher, and E. Sackmann, Microviscoelasticity of the apical cell surface of human umbilical vein endothelial cells (HUVEC) within confluent monolayers. Biophys J, 2004. 87(2): p. 1338-50.
103. Izzard, C.S. and L.R. Lochner, Cell-to-substrate contacts in living fibroblasts: an interference reflexion study with an evaluation of the technique. J Cell Sci, 1976. 21(1): p. 129-59.
156
104. Alcaraz, J., et al., Microrheology of human lung epithelial cells measured by atomic force microscopy. Biophys J, 2003. 84(3): p. 2071-9.
105. Srivastava, A. and G. Krishnamoorthy, Cell type and spatial location dependence of cytoplasmic viscosity measured by time-resolved fluorescence microscopy. Arch Biochem Biophys, 1997. 340(2): p. 159-67.
106. Bicknese, S., et al., Cytoplasmic viscosity near the cell plasma membrane: measurement by evanescent field frequency-domain microfluorimetry. Biophys J, 1993. 65(3): p. 1272-82.
107. Sato, M., et al., Application of the micropipette technique to the measurements of cultured porcine aortic endothelial cell viscoelastic properties. J. Biomech. Eng., 1990. 112: p. 263-268.
108. Desprat, N., et al., Creep function of a single living cell. Biophys J, 2005. 88(3): p. 2224-33.
109. Palmer, A., et al., Diffusing wave spectroscopy microrheology of actin filament networks. Biophys J, 1999. 76(2): p. 1063-71.
110. Li, F., J.H. Wang, and Q.M. Wang, Monitoring cell adhesion by using thickness shear mode acoustic wave sensors. Biosens Bioelectron. , 2007. 23(1): p. 42-50.
111. Harding, G.L., Mass sensitivity of Love-mode acoustic sensors incorporating silicon dioxide and silicon-oxy-fluoride guiding layers. Sensors and Actuators A, 2001. 88: p. 20-18.
112. Lee, J.Y., et al., Analysis of local tissue-specific gene expression in cellular micropatterns. Anal Chem, 2006. 78(24): p. 8305-12.
113. Langer, R. and J.P. Vacanti, Tissue engineering. Science, 1993. 260(5110): p. 920-6.
114. Mrksich, M., et al., Controlling cell attachment on contoured surfaces with self-assembled monolayers of alkanethiolates on gold. Proc Natl Acad Sci U S A, 1996. 93(20): p. 10775-8.
115. Thery, M., et al., Cell distribution of stress fibres in response to the geometry of the adhesive environment. Cell Motil Cytoskeleton, 2006. 63(6): p. 341-55.
116. Hu, S., et al., Intracellular stress tomography reveals stress focusing and structural anisotropy in cytoskeleton of living cells. Am J Physiol Cell Physiol, 2003. 285(5): p. C1082-90.
117. Chicurel, M.E., C.S. Chen, and D.E. Ingber, Cellular control lies in the balance of forces. Curr Opin Cell Biol, 1998. 10(2): p. 232-9.
157
118. Engler, A.J., et al., Myotubes differentiate optimally on substrates with tissue-like stiffness: pathological implications for soft or stiff microenvironments. J Cell Biol, 2004. 166(6): p. 877-87.
119. Li, B., et al., Development of micropost force sensor array with culture experiments for determination of cell traction forces. Cell Motil Cytoskeleton, 2007. 64(7): p. 509-18.
120. Fujimoto, H., et al., Fabrication of cell-based arrays using micropatterned alkanethiol monolayers for the parallel silencing of specific genes by small interfering RNA. Bioconjug Chem, 2006. 17(6): p. 1404-10.
121. Benjamin, M. and J.R. Ralphs, Biology of fibrocartilage cells. Int Rev Cytol, 2004. 233: p. 1-45.
122. Geiger, B. and A. Bershadsky, Exploring the neighborhood: adhesion-coupled cell mechanosensors. Cell, 2002. 110(2): p. 139-42.
123. Ingber, D.E., Tensegrity: the architectural basis of cellular mechanotransduction. Annu Rev Physiol, 1997. 59: p. 575-99.
124. Zhu, C., G. Bao, and N. Wang, Cell mechanics: mechanical response, cell adhesion, and molecular deformation. Annu Rev Biomed Eng, 2000. 2: p. 189-226.
125. Matthews, B.D., et al., Mechanical properties of individual focal adhesions probed with a magnetic microneedle. Biochem Biophys Res Commun, 2004. 313(3): p. 758-64.
126. Wang, N., et al., Micropatterning tractional forces in living cells. Cell Motil Cytoskeleton, 2002. 52(2): p. 97-106.
127. Agarwal, C., et al., Healing and normal fibroblasts exhibit differential proliferation, collagen production, alpha-SMA expression, and contraction. Ann Biomed Eng, 2006. 34(4): p. 653-9.
128. Eastwood, M., et al., Effect of precise mechanical loading on fibroblast populated collagen lattices: morphological changes. Cell Motil Cytoskeleton, 1998. 40(1): p. 13-21.
129. Huang, D., et al., Mechanisms and dynamics of mechanical strengthening in ligament-equivalent fibroblast-populated collagen matrices. Ann Biomed Eng, 1993. 21(3): p. 289-305.
158
![Soft UV Nanoimprint Lithography and Its Applications · UV-NIL, etc. Verschuuren et al. [1, 14, 18-19] proposed substrate conformal imprint lithography (SCIL), which combines the](https://static.fdocuments.in/doc/165x107/603029c202318c49852effc8/soft-uv-nanoimprint-lithography-and-its-applications-uv-nil-etc-verschuuren-et.jpg)
